eBay:iPads:: sold classification:: ebayipads_base
bdanalytics
Date: (Tue) Oct 20, 2015
Introduction:
Data: Source: Training: https://inclass.kaggle.com/c/15-071x-the-analytics-edge-summer-2015/download/eBayiPadTrain.csv
New: https://inclass.kaggle.com/c/15-071x-the-analytics-edge-summer-2015/download/eBayiPadTest.csv
Time period:
Synopsis:
Based on analysis utilizing <> techniques,
Regression results: First run:
Classification results: template: prdline.my == “Unknown” -> 296 Low.cor.X.glm: Leaderboard: 0.83458 -> Rank 288 / 1884 0.85514 newobs_tbl=[N=471, Y=327]; submit_filename=template_Final_glm_submit.csv OOB_conf_mtrx=[YN=125, NY=76]=201; max.Accuracy.OOB=0.7710; opt.prob.threshold.OOB=0.6 startprice=100.00; biddable=95.42; productline=49.22; D.T.like=29.75; D.T.use=26.32; D.T.box=21.53;
prdline: -> Worse than template prdline.my == “Unknown” -> 285 All.X.no.rnorm.rf: Leaderboard: 0.82649 newobs_tbl=[N=485, Y=313]; submit_filename=prdline_Final_rf_submit.csv OOB_conf_mtrx=[YN=119, NY=80]=199; max.Accuracy.OOB=0.8339; opt.prob.threshold.OOB=0.5 startprice=100.00; biddable=84.25; D.sum.TfIdf=7.28; D.T.use=4.26; D.T.veri=2.78; D.T.scratch=1.99; D.T.box=; D.T.like=; Low.cor.X.glm: Leaderboard: 0.81234 newobs_tbl=[N=471, Y=327]; submit_filename=prdline_Low_cor_X_glm_submit.csv OOB_conf_mtrx=[YN=125, NY=74]=199; max.Accuracy.OOB=0.8205; opt.prob.threshold.OOB=0.6 startprice=100.00; biddable=96.07; prdline.my=51.37; D.T.like=29.39; D.T.use=25.43; D.T.box=22.27; D.T.veri=; D.T.scratch=;
oobssmpl: -> Low.cor.X.glm: Leaderboard: 0.83402 newobs_tbl=[N=440, Y=358]; submit_filename=oobsmpl_Final_glm_submit OOB_conf_mtrx=[YN=114, NY=84]=198; max.Accuracy.OOB=0.7780; opt.prob.threshold.OOB=0.5 startprice=100.00; biddable=93.87; prdline.my=60.48; D.sum.TfIdf=; D.T.condition=8.69; D.T.screen=7.96; D.T.use=7.50; D.T.veri=; D.T.scratch=;
category: -> Low.cor.X.glm: Leaderboard: 0.82381 newobs_tbl=[N=470, Y=328]; submit_filename=category_Final_glm_submit OOB_conf_mtrx=[YN=119, NY=57]=176; max.Accuracy.OOB=0.8011; opt.prob.threshold.OOB=0.6 startprice=100.00; biddable=79.19; prdline.my=55.22; D.sum.TfIdf=; D.T.ipad=27.05; D.T.like=21.44; D.T.box=20.67; D.T.condition=; D.T.screen=;
dataclns: -> All.X.no.rnorm.rf: Leaderboard: 0.82211 newobs_tbl=[N=485, Y=313]; submit_filename=dataclns_Final_rf_submit OOB_conf_mtrx=[YN=104, NY=75]=179; max.Accuracy.OOB=0.7977; opt.prob.threshold.OOB=0.5 startprice.log=100.00; biddable=65.85; prdline.my=7.74; D.sum.TfIdf=; D.T.use=2.01; D.T.condition=1.87; D.T.veri=1.62; D.T.ipad=; D.T.like=; Low.cor.X.glm: Leaderboard: 0.79264 newobs_tbl=[N=460, Y=338]; submit_filename=dataclns_Low_cor_X_glm_submit OOB_conf_mtrx=[YN=113, NY=74]=187; max.Accuracy.OOB=0.7977; opt.prob.threshold.OOB=0.5 -> different from prev run of 0.6 biddable=100.00; startprice.log=91.85; prdline.my=38.34; D.sum.TfIdf=; D.T.ipad=29.92; D.T.box=27.76; D.T.work=25.79; D.T.use=; D.T.condition=;
txtterms: -> top_n = c(10) Low.cor.X.glm: Leaderboard: 0.81448 newobs_tbl=[N=442, Y=356]; submit_filename=txtterms_Final_glm_submit OOB_conf_mtrx=[YN=113, NY=69]=182; max.Accuracy.OOB=0.7943; opt.prob.threshold.OOB=0.5 biddable=100.00; startprice.log=90.11; prdline.my=37.65; D.sum.TfIdf=; D.T.ipad=28.67; D.T.work=24.90; D.T.great=21.44; # [1] “D.T.condit” “D.T.condition” “D.T.good” “D.T.ipad” “D.T.new”
# [6] “D.T.scratch” “D.T.screen” “D.T.this” “D.T.use” “D.T.work”
All.X.glm: Leaderboard: 0.81016
newobs_tbl=[N=445, Y=353]; submit_filename=txtterms_Final_glm_submit
OOB_conf_mtrx=[YN=108, NY=72]=180; max.Accuracy.OOB=0.7966;
opt.prob.threshold.OOB=0.5
biddable=100.00; startprice.log=88.24; prdline.my=33.81; D.sum.TfIdf=;
D.T.scratch=25.51; D.T.use=18.97; D.T.good=16.37; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.great” “D.T.excel” “D.T.work” “D.T.ipad”
Max.cor.Y.rpart: Leaderboard: 0.79258
newobs_tbl=[N=439, Y=359]; submit_filename=txtterms_Final_rpart_submit
OOB_conf_mtrx=[YN=105, NY=76]=181; max.Accuracy.OOB=0.7954802;
opt.prob.threshold.OOB=0.5
startprice.log=100; biddable=; prdline.my=; D.sum.TfIdf=;
D.T.scratch=; D.T.use=; D.T.good=; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
All.X.no.rnorm.rf: Leaderboard: 0.80929
newobs_tbl=[N=545, Y=253]; submit_filename=txtterms_Final_rf_submit
OOB_conf_mtrx=[YN=108, NY=61]=169; max.Accuracy.OOB=0.8090395
opt.prob.threshold.OOB=0.5
startprice.log=100.00; biddable=78.82; idseq.my=63.43; prdline.my=45.57;
D.T.use=2.76; D.T.condit=2.35; D.T.scratch=2.00; D.T.good=; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
txtclstr: All.X.no.rnorm.rf: Leaderboard: 0.79363 -> 0.79573 newobs_tbl=[N=537, Y=261]; submit_filename=txtclstr_Final_rf_submit OOB_conf_mtrx=[YN=104, NY=61]=165; max.Accuracy.OOB=0.8135593 opt.prob.threshold.OOB=0.5 startprice.log=100.00; biddable=79.99; idseq.my=64.94; prdline.my=4.14; prdline.my.clusterid=1.15; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
dupobs: All.X.no.rnorm.rf: Leaderboard: 0.79295 newobs_tbl=[N=541, Y=257]; submit_filename=dupobs_Final_rf_submit OOB_conf_mtrx=[YN=114, NY=65]=179; max.Accuracy.OOB=0.7977401 opt.prob.threshold.OOB=0.5 startprice.log=100.00; biddable=94.49; idseq.my=67.40; prdline.my=4.48; prdline.my.clusterid=1.99; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
All.X.no.rnorm.rf: Leaderboard: 0.79652
newobs_tbl=[N=523, Y=275]; submit_filename=dupobs_Final_rf_submit
OOB_conf_mtrx=[YN=114, NY=65]=179; max.Accuracy.OOB=0.7977401
opt.prob.threshold.OOB=0.5
startprice.log=100.00; biddable=94.24; idseq.my=67.92;
prdline.my=4.33; prdline.my.clusterid=2.17; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
csmmdl: All.X.no.rnorm.rf: Leaderboard: 0.79396 newobs_tbl=[N=525, Y=273]; submit_filename=csmmdl_Final_rf_submit OOB_conf_mtrx=[YN=111, NY=66]=177; max.Accuracy.OOB=0.8000000 opt.prob.threshold.OOB=0.5 startprice.log=100.00; biddable=90.30; idseq.my=67.06; prdline.my=4.40; cellular.fctr=3.57; prdline.my.clusterid=2.08;
All.Interact.X.no.rnorm.rf: Leaderboard: 0.77867 newobs_tbl=[N=564, Y=234]; submit_filename=csmmdl_Final_rf_submit OOB_conf_mtrx=[YN=120, NY=53]=173; max.Accuracy.OOB=0.8045198 opt.prob.threshold.OOB=0.5 biddable=100.00; startprice.log=93.99; idseq.my=57.30; prdline.my=9.09; cellular.fctr=3.30; prdline.my.clusterid=2.35;
All.Interact.X.no.rnorm.rf: Leaderboard: 0.77152 newobs_tbl=[N=539, Y=259]; submit_filename=csmmdl_Final_rf_submit OOB_conf_mtrx=[YN=, NY=]=; max.Accuracy.OOB=0.8011299 opt.prob.threshold.OOB=0.5 biddable=100.00; startprice.log=94.93; idseq.my=57.12; prdline.my=9.29; cellular.fctr=3.20; prdline.my.clusterid=2.50; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
All.X.glmnet:
fit_RMSE=???; OOB_RMSE=115.1247; new_RMSE=115.1247;
prdline.my.fctr=100.00; condition.fctrNew=88.53; D.npnct09.log=84.34
biddable=16.48; idseq.my=57.27;spdiff:
All.Interact.X.no.rnorm.rf: Leaderboard: 0.78218 newobs_tbl=[N=517, Y=281]; submit_filename=spdiff_Final_rf_submit OOB_conf_mtrx=[YN=121, NY=38]=159; max.Accuracy.OOB=0.8203390 opt.prob.threshold.OOB=0.6 biddable=100.00; startprice.diff=57.53; idseq.my=41.31; prdline.my=11.43; cellular.fctr=2.36; prdline.my.clusterid=1.82;
All.X.no.rnorm.rf:
fit_RMSE=92.19; OOB_RMSE=130.86; new_RMSE=130.86;
biddable=100.00; prdline.my.fctr=61.92; idseq.my=57.77;
condition.fctr=29.53; storage.fctr=11.22; color.fctr=6.69;
cellular.fctr=6.11
All.X.no.rnorm.rf: Leaderboard: 0.77443
newobs_tbl=[N=606, Y=192]; submit_filename=spdiff_Final_rf_submit
OOB_conf_mtrx=[YN=112, NY=28]=140; max.Accuracy.OOB=0.8418079
opt.prob.threshold.OOB=0.6
startprice.diff=100.00; biddable=96.53; idseq.my=38.10;
prdline.my=3.65; cellular.fctr=2.21; prdline.my.clusterid=0.91; [1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
color: All.Interact.X.glmnet: fit_RMSE=88.64520; prdline.my.fctr:D.TfIdf.sum.stem.stop.Ratio=100.00; prdline.my.fctr:condition.fctr=77.35 D.TfIdf.sum.stem.stop.Ratio=68.18 prdline.my.fctr:color.fctr=68.12 prdline.my.fctr:storage.fctr=63.32
All.X.no.rnorm.rf: Leaderboard: 0.80638
newobs_tbl=[N=550, Y=248]; submit_filename=color_Final_rf_submit
OOB_conf_mtrx=[YN=108, NY=54]=162; max.Accuracy.OOB=0.8169492
opt.prob.threshold.OOB=0.5
biddable=100.00; startprice.diff=77.90; idseq.my=48.49;
D.ratio.sum.TfIdf.nwrds=6.48; storage.fctr=4.74;
D.TfIdf.sum.stem.stop.Ratio=4.57; prdline.my=4.32;[1] “D.T.condit” “D.T.use” “D.T.scratch” “D.T.new” “D.T.good” “D.T.screen” [7] “D.T.ipad” “D.T.great” “D.T.work” “D.T.excel”
assctxt: select_terms: [1] “condit” “use” “scratch” “new” “good” “ipad” “screen” “great”
[9] “work” “excel” “like” “box” “function” “item” “fulli” “minor” [17] “cosmet” “crack” “mint” “wear”
assoc_terms: [1] “bare” “sign” “light” “back” “hous” “tab” “dent”
[8] “brand” “open” “mini” “appl” “air” “wifi” “affect”
[15] “protector” “shape” “perfect” “order” “button” “origin” “retail”
[22] “seal” “includ” “100” “may” “show” “overal” “bodi”
[29] “phone” “will” “damag” “near” “top” “normal” “tear”
[36] “expect” “minim”
glb_allobs_df\(prdline.my\).clusterid Entropy: 0.6665 (97.3037 pct) All.Interact.X.glmnet: fit_RMSE=88.40723; prdline.my.fctr:D.TfIdf.sum.stem.stop.Ratio=100.00; prdline.my.fctriPadAir:D.npnct01.log=79.67748; D.TfIdf.sum.stem.stop.Ratio=79.08192; prdline.my.fctriPadmini 2+:condition.fctrNew other (see details)=78.24020; prdline.my.fctriPad 3+:color.fctrSpace Gray=77.05886; prdline.my.fctriPadmini 2+:storage.fctrUnknown=75.68145; prdline.my.fctrUnknown:.clusterid.fctr3=74.23727;
All.Interact.X.no.rnorm.rf: Leaderboard: 0.72974
newobs_tbl=[N=682, Y=116]; submit_filename=assctxt_Final_rf_submit
OOB_conf_mtrx=[YN=125, NY=43]=168; max.Accuracy.OOB=0.8101695; max.auc.OOB=???;
opt.prob.threshold.OOB=0.6
biddable=100.00; startprice.diff=51.04; idseq.my=29.51;
startprice.diff:biddable=28.70
prdline.my.fctriPadmini:idseq.my=6.89
Highest max.auc.OOB=???; for model: ctgry2: select_terms: 50 assoc_terms: 103 glb_allobs_df\(prdline.my\).clusterid Entropy: 0.6559 (96.7556 pct) All.Interact.X.glmnet: next: All.X.glmnet fit_RMSE=88.80010; prdl.my.descr.fctr:storage.fctr 100.00 prdl.my.descr.fctr:condition.fctr 93.96 prdl.my.descr.fctr:D.npnct01.log 89.94 D.TfIdf.sum.stem.stop.Ratio 75.90 prdl.my.descr.fctr:color.fctr 72.43 prdl.my.descr.fctr:.clusterid.fctr7 63.97 prdl.my.descr.fctr:D.npnct08.log 63.46 prdl.my.descr.fctr 63.05 prdl.my.descr.fctr:D.TfIdf.sum.stem.stop.Ratio 62.91 prdl.my.descr.fctr:D.npnct16.log 62.39
Ensemble.glmnet: Leaderboard: 0.80480
newobs_tbl=[N=473, Y=325]; submit_filename=ctgry2_Final_glmnet_submit
OOB_conf_mtrx=[YN=79, NY=101]=180;
max.Accuracy.OOB=0.7977528; max.auc.OOB=0.8554068; opt.prob.threshold.OOB=0.4
Highest max.auc.OOB=0.8587215; for model:All.X.no.rnorm.rf
biddable 100.000
startprice.diff 71.793
idseq.my 43.511
ensemble: select_terms: 50 assoc_terms: 103 glb_allobs_df\(prdline.my\).clusterid Entropy: 0.6570 (96.9282 pct) Final.glment: min.RMSE.fit=31.45801 Ensemble.glmnet: min.RMSE.fit=30.67172 startprice.predict.All.Interact.X.no.rnorm.rf 100.000 startprice.predict.All.X.no.rnorm.rf 75.381 All.X.glmnet: min.RMSE.fit=88.98066 prdl.my.descr.fctr 100.00 D.TfIdf.sum.stem.stop.Ratio 92.16 condition.fctr 79.01 prdl.my.descr.fctr:.clusterid.fctr5 69.91 D.npnct16.log 61.70 color.fctrWhite 59.42 D.npnct01.log 55.07 cellular.fctr1 53.35 D.terms.n.post.stop 52.92
Ensemble.glmnet: Leaderboard: 0.73183
newobs_tbl=[N=557, Y=241]; submit_filename=ensemble_Final_glmnet_submit
OOB_conf_mtrx=[YN=75, NY=60]=135;
max.Accuracy.OOB=0.8483146; max.auc.OOB=0.9187365;
opt.prob.threshold.OOB=0.5sold.fctr.predict.All.X.no.rnorm.rf.prob 100.000000
sold.fctr.predict.All.Interact.X.no.rnorm.rf.prob 98.873608
Highest max.auc.OOB=0.9180131; for model:All.X.no.rnorm.rf
startprice.diff 100.000 biddable 95.318 idseq.my 33.365
ncv7: select_terms: 50 assoc_terms: 103 glb_allobs_df\(prdline.my\).clusterid Entropy: 0.6570 (96.9282 pct) Final.glment: min.RMSE.fit=31.45704 Ensemble.glmnet: min.RMSE.fit=29.93289 startprice.predict.All.Interact.X.no.rnorm.rf 100.000 startprice.predict.All.X.no.rnorm.rf 82.878 startprice.predict.Low.cor.X.lm 42.664
All.Interact.X.glmnet: min.RMSE.fit=87.30181 prdl.my.descr.fctr:D.npnct01.log 100.00 prdl.my.descr.fctr:condition.fctr 99.98 prdl.my.descr.fctr:storage.fctr 96.26 prdl.my.descr.fctriPadAir 79.65 prdl.my.descr.fctr:color.fctr 79.30 prdl.my.descr.fctr:D.TfIdf.sum.stem.stop.Ratio 74.90 D.TfIdf.sum.stem.stop.Ratio 74.77 prdl.my.descr.fctr:D.npnct08.log 67.38 prdl.my.descr.fctr:D.npnct01.log 67.38
Ensemble.glmnet: Leaderboard: not submitted -> lower max.auc.OOB of "Ensemble submission"
newobs_tbl=[N=561, Y=237]; submit_filename=ncv7_Final_glmnet_submit
OOB_conf_mtrx=[YN=79, NY=55]=134;
max.Accuracy.OOB=0.8494382; max.auc.OOB=0.9130918; opt.prob.threshold.OOB=0.5sold.fctr.predict.All.X.no.rnorm.rf.prob 100.000 sold.fctr.predict.All.Interact.X.no.rnorm.rf.prob 99.348
Highest max.auc.OOB=0.9117720; for model:All.X.no.rnorm.rf
startprice.diff 100.000 biddable 96.383 idseq.my 32.634
gbm w/startprice.unit9: Final.glment: min.RMSE.fit=30.32782 Ensemble.glmnet: min.RMSE.fit=29.62348 startprice.predict.All.Interact.X.no.rnorm.rf 100.000 startprice.predict.All.X.no.rnorm.rf 73.521 startprice.predict.All.Interact.X.bayesglm 29.675 startprice.predict.Max.cor.Y.lm 28.405
All.X.glmnet: min.RMSE.fit=88.64271prdl.my.descr.fctr 100.00 D.TfIdf.sum.stem.stop.Ratio 85.01 condition.fctr 80.28 carrier.fctr 77.48 prdl.my.descr.fctr:.clusterid.fctr5 65.78 D.npnct16.log 61.66 startprice.unit9 59.48 color.fctr 59.21 D.npnct01.log 53.78 D.npnct08.log 53.56 cellular.fctr 53.19
Ensemble.glmnet: Leaderboard: not submitted -> lower max.auc.OOB of "Ensemble submission"
newobs_tbl=[N=579, Y=219]; submit_filename=gbm_Final_glmnet_submit
OOB_conf_mtrx=[YN=85, NY=54]=139;
max.Accuracy.OOB=0.8438202; max.auc.OOB=0.9127314; opt.prob.threshold.OOB=0.5sold.fctr.predict.All.X.no.rnorm.rf.prob 100.0000 sold.fctr.predict.All.Interact.X.no.rnorm.rf.prob 98.7937
Highest max.auc.OOB=0.9167568; for model:All.Interact.X.gbm biddable 100.0000 startprice.diff 96.2076 startprice.diff:biddable 23.2114 idseq.my 7.8098
mdlsel: Final.glment: min.RMSE.fit=30.47114 (higher than gbm w/startprice.unit9) Ensemble.glmnet: min.RMSE.fit=29.49418 startprice.predict.All.Interact.X.no.rnorm.rf 100.000000 startprice.predict.All.X.no.rnorm.rf 71.213880 startprice.predict.All.X.bayesglm 24.166084
All.X.glmnet: min.RMSE.fit=88.64271prdl.my.descr.fctr 100.00 D.TfIdf.sum.stem.stop.Ratio 85.01 condition.fctr 80.28 carrier.fctr 77.48 prdl.my.descr.fctr:.clusterid.fctr5 65.78 D.npnct16.log 61.66 startprice.unit9 59.48 color.fctr 59.21 D.npnct01.log 53.78 D.npnct08.log 53.56 cellular.fctr 53.19
mdlsel(startprice.log): Final.Ensemble.rf: min.RMSE.fit=0.4563772 Ensemble.rf: min.RMSE.fit=0.4283013 startprice.log.predict.All.Interact.X.no.rnorm.rf 100.0000000 startprice.log.predict.All.X.no.rnorm.rf 58.0967582 startprice.log.predict.All.Interact.X.gbm 6.7197148
All.X.no.rnorm.rf: min.RMSE.fit=1.4967021biddable 100.00000000 idseq.my 98.00292371 startprice.unit9 34.31130220 prdl.my.descr.fctr 18.10984741 D.ratio.sum.TfIdf.nwrds 15.23549621 color.fctrUnknown 14.05520993 D.TfIdf.sum.stem.stop.Ratio 13.00884673 D.ratio.nstopwrds.nwrds 10.51165302
All.X.gbm: Leaderboard: 0.75430
newobs_tbl=[N=582, Y=216]; submit_filename=mdlsel_Final_gbm_submit
OOB_conf_mtrx=[YN=58, NY=65]=123;
max.Accuracy.OOB=0.8617978; max.auc.OOB=0.9367161;
opt.prob.threshold.OOB=0.5startprice.diff 100.0000000 100.00000000 biddable 66.6475055 65.40764971 idseq.my 1.8632456 4.55963698
splogdiff: All.X.gbm: Leaderboard: 0.70111 newobs_tbl=[N=553, Y=245]; submit_filename=splogdiff_Final_gbm_submit OOB_conf_mtrx=[YN=35, NY=101]=136; max.Accuracy.OOB=0.8471910; max.auc.OOB=0.9388912; opt.prob.threshold.OOB=0.3 startprice.log.diff 100.0000000 100.0000000 biddable 86.8563123 88.0261866 idseq.my 8.3580281 2.9054298
nofrcdups: All.X.gbm: Leaderboard: ???/0.67225 newobs_tbl=[N=543, Y=255]; submit_filename=nofrcdups_Final_gbm_submit OOB_conf_mtrx=[YN=36, NY=101]=137; opt.prob.threshold.OOB=0.3 max.Accuracy.OOB=0.0.8460674; max.auc.OOB=0.9388582; startprice.log.diff 100.00000000 93.3716491 biddable 83.57786348 100.0000000 idseq.my 11.54696712 1.1240259
nofrcdups w/ glb_sel_mdl_id=All.X.no.rnorm.rf: All.X.no.rnorm.rf: Leaderboard: ???/0.57475 -> ???/0.59937 with force dups newobs_tbl=[N=630, Y=168]; newobs_range_outliers=798; submit_filename=nofrcdups_Final_rf_submit OOB_conf_mtrx=[YN=70, NY=62]=132; OOBobs_range_outliers=27 opt.prob.threshold.OOB=0.5 max.Accuracy.OOB=0.8516854; max.auc.OOB=0.9335308; startprice.log.diff 100.00 biddable 87.61 idseq.my 26.79
Forum Ideas: I then focused on feature engineering, each new variable brought its own little improvement so in the end i just kept adding new ones and let the models do their thing. Here are some i used: model (productline:storage:condition), isNew, model2 (product:isNew), 50 common words from descr, descrLength, capsFactor (% of caps in description), number of cheaper items of same model2, number of dearer items of same model2, priceFactor (vs. mean of price for model), priceFactor2 (vs. mean of price for model2), bigID (if ID> 11000 because there seems to be a huge drop in sales after some time), timeline (year of product launch, reasoning is you want to spend less money on older products).
Get the median startprice for each level of productline and condition. Take the difference from startprice as a new variable. I find median works much better than the mean since startprice is not normally distributed. I also created another binary variable on whether this difference is positive or negative.
Square root startprice
scale and center all the variables except sold, including the dummies.
Prediction Accuracy Enhancement Options:
- Not repeatable experiments:
- gbm & ensembles that include gbm
- rf that includes .rnorm ??? (now .rnorm has a set seed)
- import.data chunk:
- which obs should be in fit vs. OOB (currently dirty.0 vs .1 is split 50%)
- inspect.data chunk:
- For date variables
- Appropriate factors ?
- Different / More last* features ?
- For date variables
- scrub.data chunk:
- transform.data chunk:
- derive features from multiple features
- manage.missing.data chunk:
- Not fill missing vars
- Fill missing numerics with a different algorithm
- Fill missing chars with data based on clusters
- extract.features chunk:
- Text variables: move to date extraction chunk ???
- Mine acronyms
- Mine places
- Text variables: move to date extraction chunk ???
- Review set_global_options chunk after features are finalized
[](.png)
Potential next steps include:
- Organization:
- Categorize by chunk
- Priority criteria:
- Ease of change
- Impacts report
- Cleans innards
- Bug report
- all chunks:
- at chunk-end rm(!glb_)
- manage.missing.data chunk:
- cleaner way to manage re-splitting of training vs. new entity
- extract.features chunk:
- Add n-grams for glb_txt_vars
- “RTextTools”, “tau”, “RWeka”, and “textcat” packages
- Convert user-specified mutate code to config specs
- Add n-grams for glb_txt_vars
- fit.models chunk:
- Prediction accuracy scatter graph:
- Add tiles (raw vs. PCA)
- Use shiny for drop-down of “important” features
Use plot.ly for interactive plots ?
- Change .fit suffix of model metrics to .mdl if it’s data independent (e.g. AIC, Adj.R.Squared - is it truly data independent ?, etc.)
- move model_type parameter to myfit_mdl before indep_vars_vctr (keep all model_* together)
- create a custom model for rpart that has minbucket as a tuning parameter
varImp for randomForest crashes in caret version:6.0.41 -> submit bug report
- Probability handling for multinomials vs. desired binomial outcome
- ROCR currently supports only evaluation of binary classification tasks (version 1.0.7)
extensions toward multiclass classification are scheduled for the next release
- Skip trControl.method=“cv” for dummy classifier ?
- Add custom model to caret for a dummy (baseline) classifier (binomial & multinomial) that generates proba/outcomes which mimics the freq distribution of glb_rsp_var values; Right now glb_dmy_glm_mdl always generates most frequent outcome in training data
glm_dmy_mdl should use the same method as glm_sel_mdl until custom dummy classifer is implemented
- fit.all.training chunk:
- myplot_prediction_classification: displays ‘x’ instead of ‘+’ when there are no prediction errors
- Compare glb_sel_mdl vs. glb_fin_mdl:
- varImp
- Prediction differences (shd be minimal ?)
- Move glb_analytics_diag_plots to mydsutils.R: (+) Easier to debug (-) Too many glb vars used
- Add print(ggplot.petrinet(glb_analytics_pn) + coord_flip()) at the end of every major chunk
- Parameterize glb_analytics_pn
- Move glb_impute_missing_data to mydsutils.R: (-) Too many glb vars used; glb_<>_df reassigned
- Replicate myfit_mdl_classification features in myfit_mdl_regression
- Do non-glm methods handle interaction terms ?
- f-score computation for classifiers should be summation across outcomes (not just the desired one ?)
- Add accuracy computation to glb_dmy_mdl in predict.data.new chunk
- Why does splitting fit.data.training.all chunk into separate chunks add an overhead of ~30 secs ? It’s not rbind b/c other chunks have lower elapsed time. Is it the number of plots ?
- Incorporate code chunks in print_sessionInfo
- Test against
- projects in github.com/bdanalytics
- lectures in jhu-datascience track
Analysis:
rm(list = ls())
set.seed(12345)
options(stringsAsFactors = FALSE)
source("~/Dropbox/datascience/R/myscript.R")
source("~/Dropbox/datascience/R/mydsutils.R")## Loading required package: caret
## Loading required package: lattice
## Loading required package: ggplot2source("~/Dropbox/datascience/R/myplot.R")
source("~/Dropbox/datascience/R/mypetrinet.R")
source("~/Dropbox/datascience/R/myplclust.R")
source("~/Dropbox/datascience/R/mytm.R")
# Gather all package requirements here
suppressPackageStartupMessages(require(doMC))
registerDoMC(6) # # of cores on machine - 2
suppressPackageStartupMessages(require(caret))
source("~/Documents/Work/PullRequests/caret/pkg/caret/R/confusionMatrix.R")
source("~/Documents/Work/PullRequests/caret/pkg/caret/R/ggplot.R")
#packageVersion("tm")
#require(sos); findFn("cosine", maxPages=2, sortby="MaxScore")
# Analysis control global variables
glb_trnng_url <- "https://inclass.kaggle.com/c/15-071x-the-analytics-edge-summer-2015/download/eBayiPadTrain.csv"
glb_newdt_url <- "https://inclass.kaggle.com/c/15-071x-the-analytics-edge-summer-2015/download/eBayiPadTest.csv"
glb_out_pfx <- "ebayipads_base_"
glb_save_envir <- FALSE # or TRUE
glb_is_separate_newobs_dataset <- TRUE # or TRUE
glb_split_entity_newobs_datasets <- FALSE # select from c(FALSE, TRUE)
glb_split_newdata_method <- NULL # select from c(NULL, "condition", "sample", "copy")
glb_split_newdata_condition <- NULL # or "is.na(<var>)"; "<var> <condition_operator> <value>"
glb_split_newdata_size_ratio <- 0.3 # > 0 & < 1
glb_split_sample.seed <- 123 # or any integer
glb_obs_drop_condition <- # default : NULL
"(UniqueID %in% c(NULL
, 11234 #sold=0; 2 other dups(10306, 11503) are sold=1
, 11844 #sold=0; 3 other dups(11721, 11738, 11812) are sold=1
))
# | (productline %in% c('iPad 5', 'iPad mini Retina'))
# | (biddable != 0) # bid0_sp
# | (biddable == 0) # bid1_sp
"
#parse(text=glb_obs_drop_condition)
#subset(glb_allobs_df, .grpid %in% c(31))
glb_obs_repartition_train_condition <- NULL
# "!is.na(sold) & (sold == 1)" # : bid._sp
glb_max_fitobs <- NULL # or any integer
glb_is_regression <- FALSE; glb_is_classification <- !glb_is_regression;
glb_is_binomial <- TRUE #or FALSE
glb_rsp_var_raw <- "sold" #: !_sp # "startprice" # : bid._sp #
# for classification, the response variable has to be a factor
glb_rsp_var <- "sold.fctr" #:!_sp # "startprice.log10" :bid._sp # glb_rsp_var_raw :default
# if the response factor is based on numbers/logicals e.g (0/1 OR TRUE/FALSE vs. "A"/"B"),
# or contains spaces (e.g. "Not in Labor Force")
# caret predict(..., type="prob") crashes
glb_map_rsp_raw_to_var <- function(raw) { # NULL
# return(raw ^ 0.5)
# return(log(1 + raw))
# return(log10(raw)) # bid._sp
# return(exp(-raw / 2))
ret_vals <- rep_len(NA, length(raw)); ret_vals[!is.na(raw)] <- ifelse(raw[!is.na(raw)] == 1, "Y", "N"); return(relevel(as.factor(ret_vals), ref="N"))
# #as.factor(paste0("B", raw))
# #as.factor(gsub(" ", "\\.", raw))
}
# glb_map_rsp_raw_to_var(tst <- c(NA, 0, 1)) # !_sp
# glb_map_rsp_raw_to_var(tst <- c(NA, 0, 2.99, 280.50, 1000.00)) # bid._sp
glb_map_rsp_var_to_raw <- function(var) { # NULL #
# return(var ^ 2.0)
# return(exp(var) - 1)
# return(10 ^ var) # bid._sp
# return(-log(var) * 2)
as.numeric(var) - 1
# #as.numeric(var)
# #gsub("\\.", " ", levels(var)[as.numeric(var)])
# c("<=50K", " >50K")[as.numeric(var)]
# #c(FALSE, TRUE)[as.numeric(var)]
}
# glb_map_rsp_var_to_raw(glb_map_rsp_raw_to_var(tst))
if ((glb_rsp_var != glb_rsp_var_raw) && is.null(glb_map_rsp_raw_to_var))
stop("glb_map_rsp_raw_to_var function expected")
glb_rsp_var_out <- paste0(glb_rsp_var, ".predict.") # mdl_id is appended later
# List info gathered for various columns
# <col_name>: <description>; <notes>
# description = The text description of the product provided by the seller.
# biddable = Whether this is an auction (biddable=1) or a sale with a fixed price (biddable=0).
# startprice = The start price (in US Dollars) for the auction (if biddable=1) or the sale price (if biddable=0).
# condition = The condition of the product (new, used, etc.)
# cellular = Whether the iPad has cellular connectivity (cellular=1) or not (cellular=0).
# carrier = The cellular carrier for which the iPad is equipped (if cellular=1); listed as "None" if cellular=0.
# color = The color of the iPad.
# storage = The iPad's storage capacity (in gigabytes).
# productline = The name of the product being sold.
# If multiple vars are parts of id, consider concatenating them to create one id var
# If glb_id_var == NULL, ".rownames <- row.names()" is the default
# Derive a numeric feature from id var
# User-specified exclusions
# List feats that shd be excluded due to known causation by prediction variable
glb_exclude_vars_as_features <- c(NULL
### !_sp
, "description", "descr.my", "productline"
### bid0_sp
# , "description", "productline"
# , "sold", "startprice.log10.cut.fctr"
# # List feats that are linear combinations (alias in glm)
# , "D.terms.post.stem.n.log", "D.weight.sum"
# #, "prdl.descr.my.fctriPad4#1:.clusterid.fctr3" This does not work
# # if RFE is rated lower than Low.cor, list feats that are in RFE & not in Low.cor
# # min.RMSE.fit(RFE.X.glmnet)=0.1138888
# # D.chrs.n.log 61.12483
# # D.chrs.uppr.n.log 61.12483
# # D.ratio.wrds.stop.n.wrds.n 61.12483
# # D.terms.post.stop.n.log 61.12483
# # D.weight.post.stem.sum 61.12483
# # D.wrds.n.log 61.12483
# # D.wrds.stop.n.log 61.12483
# # D.wrds.unq.n.log 61.12483
# #, "startprice.dcm2.is9" # min.RMSE.fit(RFE.X.glmnet)=0.1141991 (up)
# , "D.wrds.stop.n.log" # min.RMSE.fit(RFE.X.glmnet)=0.1131232
### bid0_sp
### bid1_sp
# , "description", "productline"
# , "sold", "startprice.log10.cut.fctr"
### bid1_sp
)
glb_id_var <- c("UniqueID")
glb_category_var <- "prdl.descr.my.fctr" # "productline" # NULL
glb_drop_vars <- c(NULL) # or c("<col_name>")
glb_map_vars <- NULL # or c("<var1>", "<var2>")
glb_map_urls <- list();
# glb_map_urls[["<var1>"]] <- "<var1.url>"
glb_assign_pairs_lst <- NULL;
# glb_assign_pairs_lst[["<var1>"]] <- list(from=c(NA),
# to=c("NA.my"))
glb_assign_vars <- names(glb_assign_pairs_lst)
# Derived features
glb_derive_lst <- NULL;
# Add logs of numerics that are not distributed normally -> do automatically ???
# Right skew: logp1; sqrt; ^ 1/3; logp1(logp1); log10; exp(-<feat>/constant)
# glb_derive_lst[["prdline.my"]] <- list(
# mapfn=function(productline) { return(productline) }
# , args=c("productline"))
### bid._sp
# glb_derive_lst[["startprice.log10.cut.fctr"]] <- list(
# mapfn=function(startprice.log10) { return(cut(startprice.log10, 3)) }
# , args=c("startprice.log10"))
### bid._sp
glb_derive_lst[["descr.my"]] <- list(
mapfn=function(description) { mod_raw <- description;
### bid._sp
# # This is here because it does not work with txt_map_filename
# mod_raw <- gsub(paste0(c("\n", "\211", "\235", "\317", "\333"), collapse="|"), " ",
# mod_raw)
# # This should go into txt_map_filename
# mod_raw <- gsub("\\.\\.", "\\. ", mod_raw);
# # Don't parse for "." because of ".com"; use customized gsub for that text
# mod_raw <- gsub("(\\w)(!|\\*|,|-|/)(\\w)", "\\1\\2 \\3", mod_raw);
#
# # Modifications for this exercise only
# # Add dictionary to stemDocument e.g. stickers stemmed to sticker ???
# mod_raw <- gsub("8\\.25", "825", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" 10\\.SCREEN ", " 10\\. SCREEN ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" 128 gb ", " 128gb ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" 16G, ", " 16GB, ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" 16 gig ", " 16gb ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" 16 gb ", " 16gb ", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub("\\bAccounts\\b", "Account", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bactivated\\b", "activate", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" actuuly ", " actual ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" Apple care ", " Applecare ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\baffects\\b", "affect", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bApple\\'s", "Apple", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(665), txt_var]; mod_raw
# mod_raw <- gsub(" Apple care ", " Applecare ", mod_raw, ignore.case=FALSE);
# ub(" bacK!wiped ", " bacK ! wiped ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" backplate", " back plate", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bbarley", "barely", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" bend ", " bent ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("Best Buy", "BestBuy", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" black\\.Device ", " black \\. Device ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("black\\),charger ", "black\\), charger ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("\\bblacked\\b", "black", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bblemish\\b", "blemishes", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" blocks", " blocked", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" brokenCharger ", " broken Charger ", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub(" perfectlycord ", " perfectly cord ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bcord", "cable", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bcables\\b", "cable", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub(" care\\.The ", " care\\. The ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(cared|careful|CAREFUL)\\b", "care", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\b(cases|casing)\\b", "case", mod_raw, ignore.case=TRUE);
# #mod_raw <- glb_allobs_df[c(88,187,280,1040,1098), txt_var]; mod_raw
# mod_raw <- gsub(" carefully ", " careful ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bchargers\\b", "charger", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bchips\\b", "chip", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bcleanly\\b", "clean", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\b(C|c)olor(.*)s\\b", "\\1olor", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(280,1411), txt_var]; mod_raw
# mod_raw <- gsub("\\bcompletely\\b", "complete", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(42,502,676), txt_var]; mod_raw
#
# mod_raw <- gsub(" (conditon|condtion|contidion|conditions)", " condition", mod_raw,
# mod_raw <- gsub("\\b(conditon|condtion|contidion|conditions)\\b", "condition", mod_raw,
# ", "\\1\\. \\2", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("(condition)(Has)", "\\1\\. \\2", mod_raw);
#
# mod_raw <- gsub("\\bCONNECTED\\b", "CONNECT", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bconnects\\b", "connect", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" consist ", " consistent ", mod_raw, ignore.case=TRUE);
# #mod_raw <- glb_allobs_df[c(195, 379, 437), txt_var]; mod_raw
# mod_raw <- gsub("\\bCosmetics\\b", "Cosmetic", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" cracksNo ", " cracks No ", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub("\\b(D|d)amaged\\b", "\\1amage", mod_raw, ignore.case=TRUE);
# #mod_raw <- glb_allobs_df[c(116, 1360), txt_var]; mod_raw
# mod_raw <- gsub("\\bDays\\b", "Day", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" DEFAULTING ", " DEFAULT ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bdefect(ive)*\\b", "defects", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(1403), txt_var]; mod_raw
# mod_raw <- gsub(" definitely ", " definite ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(D|d)ented\\b", "\\1ent", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" described", " describe", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" desciption", " description", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" devices", " device", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" Digi\\.", " Digitizer\\.", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(ding|dinged)\\b", "dings", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" display\\.New ", " display\\. New ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" displays", " display", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(drop|drops)\\b", "dropped", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(E|e)dge\\b", "\\1dges", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" effect ", " affect ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" Excellant ", " Excellent ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" excellently", " excellent", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" EUC ", " excellent used condition", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" feels ", " feel ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" fineiCloud ", " fine iCloud ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" fine.Its ", " fine. Its ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bfix\\b", "fixed", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bflaws\\b", "flaw", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bflawlessly\\b", "flawless", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" Framing ", " Frame ", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub(" functioanlity", " functionality", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bfunction(ing|ality)\\b", "functional", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" functional\\.Very little ", " functional\\. Very little ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("^Gentle ", "Gently ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\(gray color", "\\(spacegray color", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" GREAT\\.SCreen ", " GREAT\\. SCreen ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("\\b(guarantee|guarantees)\\b", "guaranteed", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\ba handful of times\\b", "sparingly", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bhardly any\\b", "no", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bhardly ever used\\b", "sparingly used", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub("iCL0UD", "iCLOUD", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("----I cloud ", " ----Icloud ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("^iPad Black 3rd generation ", "iPad 3 Black ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub(" IMEINo ", " IMEI No ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bincluding\\b", "included", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" install\\. ", " installed\\. ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("inivisible", "invisible", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("^I pad ", "Ipad ", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub("\\b(lock|locks)\\b", "locked", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\blots\\b", "lot", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" manuals ", " manual ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" book ", " manual ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" mars ", " marks ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" marks\\.Absolutely ", " marks\\. Absolutely ", mod_raw,
# ignore.case=TRUE);
# mod_raw <- gsub("\\bmarkings\\b", "marks", mod_raw, ignore.case=TRUE);
# #mod_raw <- glb_allobs_df[c(200, 1301), txt_var]; mod_raw
# mod_raw <- gsub(" minimum", " minimal", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" MINT\\.wiped ", " MINT\\. wiped ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bmonth\\b", "months", mod_raw, ignore.case=TRUE);
# #mod_raw <- glb_allobs_df[c(29, 38, 194, 511, 789, 819), txt_var]; mod_raw
# mod_raw <- gsub(" NEW\\!(SCREEN|ONE) ", " NEW\\! \\1 ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" new looking$", " looks new", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" newer ", " new ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bnoted\\b", "note", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" oped ", " opened ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" opening", " opened", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" operated", " operational", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("otter box", "otterbox", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub("\\bpackage\\b", "packaging", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bPACKAGE\\b", "PACKAGing", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(360, 1142), txt_var]; mod_raw
# mod_raw <- gsub(" performance", " performs", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" personalized ", " personal ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bPhysically\\b", "Physical", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\b(picture|pictured)\\b", "pictures", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bPICTURE\\b", "PICTUREs", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(184, 892), txt_var]; mod_raw
# mod_raw <- gsub("\\b[P|p]ower(ed|ing|s)\\b", "\\1ower", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" pre- owned ", " used ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bprevious\\b", "previously", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bpreviously (owned|used)\\b", "used", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bproblem\\b", "problems", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" products ", " product ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bprotected\\b", "protector", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bprotection\\b", "protector", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bPROTECTION\\b", "PROTECTOR", mod_raw, ignore.case=FALSE);
#
# mod_raw <- gsub(" Keeped ", " Kept ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" knicks ", " nicks ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" lightening ", " lightning ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bminis\\b", "mini", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("^READiPad ", "READ iPad ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" re- assemble ", " reassemble ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" REFURB\\.", " REFURBISHED\\.", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" reponding", " respond", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" rotation ", " rotate ", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub(" Sales ", " Sale ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" scratchs ", " scratches ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(scratchs|scratching)\\b", "scratches", mod_raw, ignore.case=FALSE);
# aw <- gsub(" shipped| Shipment", " ship", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bshowing\\b", "shows", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("shrink wrap", "shrinkwrap", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bshuts\\b", "shut", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" sides ", " side ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" skinned,", " skin,", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bSlightly\\b", "slight", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bspace (grey|gray)", "spacegray", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" spec ", " speck ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("^somescratches ", "some scratches ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" Sticker ", " Stickers ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bstoring", "store", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("SWAPPA\\.COM", "SWAPPAsdotCOM", mod_raw, ignore.case=TRUE);
#
# mod_raw <- gsub(" T- Mobile", " TMobile", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\b(tear|TEAR)(s|S)\\b", "\\1", mod_raw, ignore.case=FALSE);
# #mod_raw <- glb_allobs_df[c(376), txt_var]; mod_raw
# mod_raw <- gsub(" touchscreen ", " touch screen ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bTURN\\b", "TURNS", mod_raw, ignore.case=FALSE);
#
# mod_raw <- gsub(" UnlockedCracked ", " Unlocked Cracked ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bUNUSABLE\\b", "UNUSED", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\b(update|updates)\\b", "updated", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub("\\bupgrade\\b", "upgraded", mod_raw, ignore.case=FALSE);
# mod_raw <- gsub(" uppser ", " upper ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" use\\.Scratches ", " use\\. Scratches ", mod_raw,
# ignore.case=TRUE);
#
# mod_raw <- gsub(" verify ", " verified ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" wear\\.Device ", " wear\\. Device ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub("\\bwears\\b", "\\wear", mod_raw, ignore.case=TRUE);
# #mod_raw <- glb_allobs_df[c(167, 272), txt_var]; mod_raw
# mod_raw <- gsub(" whats ", " what's ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" WiFi\\+4G ", " WiFi \\+ 4G ", mod_raw, ignore.case=TRUE);
# mod_raw <- gsub(" Zaag Invisible Shield", " Zaag InvisibleShield", mod_raw,
# ignore.case=TRUE);
### bid._sp
return(mod_raw) }
, args = c("description"))
glb_derive_lst[["prdl.descr.my.fctr"]] <- list(
mapfn = function(productline, description) {
as.factor(paste(gsub(" ", "", productline), as.numeric(nchar(description) > 0),
sep = "#")) }
, args = c("productline", "description"))
#print(mycreate_sqlxtab_df(glb_allobs_df, c("prdl.descr.my.fctr", "sold")))
# mapfn=function(startprice) { return(scale(log(startprice))) }
# , args=c("startprice"))
# mapfn=function(Rasmussen) { return(ifelse(sign(Rasmussen) >= 0, 1, 0)) }
# mapfn=function(PropR) { return(as.factor(ifelse(PropR >= 0.5, "Y", "N"))) }
# mapfn=function(purpose) { return(relevel(as.factor(purpose), ref="all_other")) }
# mapfn=function(Week) { return(substr(Week, 1, 10)) }
# mapfn=function(raw) { tfr_raw <- as.character(cut(raw, 5));
# tfr_raw[is.na(tfr_raw)] <- "NA.my";
# return(as.factor(tfr_raw)) }
# , args=c("raw"))
# mapfn=function(PTS, oppPTS) { return(PTS - oppPTS) }
# , args=c("PTS", "oppPTS"))
# # If glb_allobs_df is not sorted in the desired manner
# mapfn=function(Week) { return(coredata(lag(zoo(orderBy(~Week, glb_allobs_df)$ILI), -2, na.pad=TRUE))) }
# mapfn=function(ILI) { return(coredata(lag(zoo(ILI), -2, na.pad=TRUE))) }
# mapfn=function(ILI.2.lag) { return(log(ILI.2.lag)) }
# glb_derive_lst[["<txt_var>.niso8859.log"]] <- list(
# mapfn=function(<txt_var>) { match_lst <- gregexpr("&#[[:digit:]]{3};", <txt_var>)
# match_num_vctr <- unlist(lapply(match_lst,
# function(elem) length(elem)))
# return(log(1 + match_num_vctr)) }
# , args=c("<txt_var>"))
# mapfn=function(raw) { mod_raw <- raw;
# mod_raw <- gsub("&#[[:digit:]]{3};", " ", mod_raw);
# # Modifications for this exercise only
# mod_raw <- gsub("\\bgoodIn ", "good In", mod_raw);
# return(mod_raw)
# # Create user-specified pattern vectors
# #sum(mycount_pattern_occ("Metropolitan Diary:", glb_allobs_df$Abstract) > 0)
# if (txt_var %in% c("Snippet", "Abstract")) {
# txt_X_df[, paste0(txt_var_pfx, ".P.metropolitan.diary.colon")] <-
# as.integer(0 + mycount_pattern_occ("Metropolitan Diary:",
# glb_allobs_df[, txt_var]))
#summary(glb_allobs_df[ ,grep("P.on.this.day", names(glb_allobs_df), value=TRUE)])
# glb_allobs_df$<descriptor>.my <-
# plyr::revalue(glb_allobs_df$<descriptor>.my, c(
# "ABANDONED BUILDING" = "OTHER",
# "##" = "##"
# ))
# print(<descriptor>_freq_df <- mycreate_sqlxtab_df(glb_allobs_df, c("<descriptor>.my")))
# # print(dplyr::filter(<descriptor>_freq_df, grepl("(MEDICAL|DENTAL|OFFICE)", <descriptor>.my)))
# # print(dplyr::filter(dplyr::select(glb_allobs_df, -<var.zoo>),
# # grepl("STORE", <descriptor>.my)))
# glb_derive_lst[["<var1>"]] <- glb_derive_lst[["<var2>"]]
glb_derive_vars <- names(glb_derive_lst)
# tst <- "descr.my"; args_lst <- NULL; for (arg in glb_derive_lst[[tst]]$args) args_lst[[arg]] <- glb_allobs_df[, arg]; print(head(args_lst[[arg]])); print(head(drv_vals <- do.call(glb_derive_lst[[tst]]$mapfn, args_lst)));
# print(which_ix <- which(args_lst[[arg]] == 0.75)); print(drv_vals[which_ix]);
glb_date_vars <- NULL # or c("<date_var>")
glb_date_fmts <- list(); #glb_date_fmts[["<date_var>"]] <- "%m/%e/%y"
glb_date_tzs <- list(); #glb_date_tzs[["<date_var>"]] <- "America/New_York"
#grep("America/New", OlsonNames(), value=TRUE)
glb_price_vars <- NULL # c("startprice") : bid._sp # NULL or c("<price_var>")
# Text Processing Step: custom modifications not present in txt_munge
glb_txt_vars <- NULL # c("descr.my") # NULL #
Sys.setlocale("LC_ALL", "C") # For english## [1] "C/C/C/C/C/en_US.UTF-8"# Text Processing Step: universal modifications
glb_txt_munge_filenames_pfx <- "ebay_mytxt_"
# Text Processing Step: tolower
# Text Processing Step: removePunctuation (use custom transformer to replace with space ???)
# Text Processing Step: removeWords
glb_txt_stop_words <- list()
# Remember to use unstemmed words; Check stemming of "significant" words - any stopped words that should be stemmed with them ?
if (!is.null(glb_txt_vars)) {
require(tm)
### bid._sp
# glb_txt_stop_words[["descr.my"]] <- sort(c(NULL
# , setdiff(removePunctuation(stopwords("english")), "no")
# ,"ac"
# # cor.y.train == NA
# ,unlist(strsplit(paste(c(NULL
# ,"128gb,1st,32gb,3g,64gb,90,acceptable,activation,amount,average,bad,buttons,buy,came,camera,can,care,carrier"
# #,casing
# ,"certified,charge,charging,cleaned,clear,come,components,contain,corner,correctly,covered,customer,earbuds"
# ,"engraved,engraving,engravement" # somehow didn't show up in the cor.y.train == NA list
# ,"entire,except,fair,features,feel,fine,generation,get,gift,got,heavily,heavy,however,imei,include,inspected,invisible,invisibleshield"
# ,"ipad,ipads"
# ,"issues"
# #,items,
# ,"keyboard,lightning,listing,little,looks,lower"
# ,"manufacture,manufacturer"# somehow didn't show up in the cor.y.train == NA list
# ,"meaning,model,near,need,needs,nicks,opened,operational,otherwise"
# ,"person,personal"# somehow didn't show up in the cor.y.train == NA list
# ,"phone,photos,pics,plastic,port,professionally"
# ,"purchased,purchasing"# somehow didn't show up in the cor.y.train == NA list
# ,"quality,questions,read,ready"
# ,"receive,received"# somehow didn't show up in the cor.y.train == NA list
# ,"removed,replaced,retail,return,returns,runs"
# #,scratch,
# ,"scuffing,sealed,sell,seller,selling,shape,ship,shown,silver,since,sold,sound,spacegray,stock,sync,tablet,taken,technician,tests,third,time,touch,units,unlocked,week,wifi,without"
# ,"wrap" # somehow didn't show up in the cor.y.train == NA list
# ,"zagg"
# ), collapse=",")
# , "[,]")) #err.abs.fit.sum=26.869473 w/o items,scratch
#
# # cor.y.abs is low
# #,"always","comes","grade","moderate","protector"
# ))
### bid._sp
}
# To identify terms with cor.y == NA
#orderBy(~-freq+term, subset(glb_post_stop_words_terms_df_lst[[txt_var]], is.na(cor.y)))
#paste(sort(subset(glb_post_stop_words_terms_df_lst[[txt_var]], is.na(cor.y))[, "term"]), collapse=",")
#orderBy(~-freq+term, subset(glb_post_stem_words_terms_df_lst[[txt_var]], is.na(cor.y)))
# To identify terms with low cor.y.abs
#head(orderBy(~cor.y.abs+term, subset(glb_post_stem_words_terms_df_lst[[txt_var]], !is.na(cor.y))), 5)
# To identify obs with a txt term
#tail(orderBy(~-freq+term, glb_post_stop_words_terms_df_lst[[txt_var]]), 20)
#mydsp_obs(list(descr.my.contains="non"), cols=c("color", "carrier", "cellular", "storage"))
#grep("ever", dimnames(terms_stop_mtrx)$Terms)
#which(terms_stop_mtrx[, grep("ipad", dimnames(terms_stop_mtrx)$Terms)] > 0)
#glb_allobs_df[which(terms_stop_mtrx[, grep("16", dimnames(terms_stop_mtrx)$Terms)[1]] > 0), c(glb_category_var, "storage", txt_var)]
# To identify whether terms shd be synonyms
#orderBy(~term, glb_post_stop_words_terms_df_lst[[txt_var]][grep("^moder", glb_post_stop_words_terms_df_lst[[txt_var]]$term), ])
# term_row_df <- glb_post_stop_words_terms_df_lst[[txt_var]][grep("^came$", glb_post_stop_words_terms_df_lst[[txt_var]]$term), ]
#
# cor(glb_post_stop_words_terms_mtrx_lst[[txt_var]][glb_allobs_df$.lcn == "Fit", term_row_df$pos], glb_trnobs_df[, glb_rsp_var], use="pairwise.complete.obs")
# To identify which stopped words are "close" to a txt term
#sort(cluster_vars)
# Text Processing Step: stemDocument
# To identify stemmed txt terms
#glb_post_stop_words_terms_df_lst[[txt_var]][grep("condit", glb_post_stop_words_terms_df_lst[[txt_var]]$term), ]
#orderBy(~term, glb_post_stem_words_terms_df_lst[[txt_var]][grep("^p", glb_post_stem_words_terms_df_lst[[txt_var]]$term), ])
#glb_allobs_df[which(terms_stem_mtrx[, grep("use", dimnames(terms_stem_mtrx)$Terms)[[1]]] > 0), c(glb_id_var, "productline", txt_var)]
#glb_allobs_df[which(TfIdf_stem_mtrx[, 191] > 0), c(glb_id_var, glb_category_var, txt_var)]
#which(glb_allobs_df$UniqueID %in% c(11915, 11926, 12198))
# Text Processing Step: mycombineSynonyms
# To identify which synonyms should be combined
#orderBy(~term, glb_post_stem_words_terms_df_lst[[txt_var]][grep("^c", glb_post_stem_words_terms_df_lst[[txt_var]]$term), ])
chk_comb_cor <- function(syn_lst) {
# cor(terms_stem_mtrx[glb_allobs_df$.src == "Train", grep("^(damag|dent|ding)$", dimnames(terms_stem_mtrx)[[2]])], glb_trnobs_df[, glb_rsp_var], use="pairwise.complete.obs")
print(subset(glb_post_stem_words_terms_df_lst[[txt_var]], term %in% syn_lst$syns))
print(subset(get_corpus_terms(tm_map(glb_txt_corpus_lst[[txt_var]], mycombineSynonyms, list(syn_lst), lazy=FALSE)), term == syn_lst$word))
# cor(terms_stop_mtrx[glb_allobs_df$.src == "Train", grep("^(damage|dent|ding)$", dimnames(terms_stop_mtrx)[[2]])], glb_trnobs_df[, glb_rsp_var], use="pairwise.complete.obs")
# cor(rowSums(terms_stop_mtrx[glb_allobs_df$.src == "Train", grep("^(damage|dent|ding)$", dimnames(terms_stop_mtrx)[[2]])]), glb_trnobs_df[, glb_rsp_var], use="pairwise.complete.obs")
}
#chk_comb_cor(syn_lst=list(word="cabl", syns=c("cabl", "cord")))
#chk_comb_cor(syn_lst=list(word="damag", syns=c("damag", "dent", "ding")))
#chk_comb_cor(syn_lst=list(word="dent", syns=c("dent", "ding")))
#chk_comb_cor(syn_lst=list(word="use", syns=c("use", "usag")))
glb_txt_synonyms <- list()
glb_txt_synonyms[["descr.my"]] <- NULL #: default
### bid._sp
# list(NULL
# , list(word="cabl", syns=c("cabl", "cord"))#err.abs.fit.sum=26.863220
# # , list(word="charger", syns=c("charg", "charger"))
# # , list(word="come", syns=c("came", "come"))
# # , list(word="dent", syns=c("dent", "ding"))
# # , list(word="damag", syns=c(#"bad", "blemish", "broken", "crack",
# # #defect has +ve cor, others have -ve cor
# # "damag", "dent", "ding",
# # #"scratch", "scuff", "tear", "wear",
# # NULL))
# # # combining damag with defect & dent results in higher err.abs.fit.sum=26.885899
# # # combining defect with dent in higher err.abs.fit.sum=26.894976
# # , list(word="defect", syns=c(#"bad", "blemish", "broken", "crack",
# # "defect", "dent", #"ding", ding has -ve cor, others have +ve cor
# # #"scratch", "scuff", "tear", "wear",
# # NULL))
# #, list(word="new", syns=c("brand")) ???
# # , list(word="scuff", syns=c("scuf", "scuff"))
# # , list(word="show", syns=c("show", "shown"))
# # , list(word="tablet", syns=c("tab", "tablet"))
# )
### bid._sp
if (length(glb_txt_synonyms) > 0) names(glb_txt_synonyms) <- glb_txt_vars
# Text Processing Step: filterTerms
if (!is.null(glb_txt_vars)) {
require(tm)
# options include: weightTf, myweightTflog1p, myweightTfsqrt, weightTfIdf, weightBM25
glb_txt_terms_control <- list(weighting = weightTfIdf # : default
# termFreq selection criteria across obs: default: list(global=c(1, Inf))
, bounds = list(global = c(1, Inf)) # bid._sp: list(global=c(3, Inf))
# default: c(3, Inf)
, wordLengths = c(3, Inf) # bid._sp: c(2, Inf)
)
}
glb_txt_cor_var <- glb_rsp_var # bid._sp: "startprice.log10.cut.fctr" # default: glb_rsp_var
# select one from c("union.top.val.cor", "top.cor", default: "top.val", "sparse")
glb_txt_terms_filter <- "top.val"
glb_txt_top_n <- c(10) # bid._sp: c(20) # c(50) in (old) !_sp # default: rep(10, length(glb_txt_vars))
names(glb_txt_top_n) <- glb_txt_vars
# Text Processing Step: extractAssoc
glb_txt_assoc_cor <- c(1) #bid._sp: c(0.4) #(old) !_sp: 0.2 #default: rep(1, length(glb_txt_vars))
names(glb_txt_assoc_cor) <- glb_txt_vars
# Text Processing Step: extractPatterns (ngrams)
# Potential Enhancements
# "Seller refurbished" -> D.P.refurbished.seller ?
# "Like new" -> D.P.new.like ?
# "No scratches" -> D.P.scratch.no ?
glb_important_terms <- list()
# Remember to use stemmed terms
# Have to set it even if it is not used
glb_sprs_thresholds <- c(0.950) # Generates 8 terms
# Properties:
# numrows(glb_feats_df) << numrows(glb_fitobs_df)
# Select terms that appear in at least 0.2 * O(FP/FN(glb_OOBobs_df))
# numrows(glb_OOBobs_df) = 1.1 * numrows(glb_newobs_df)
names(glb_sprs_thresholds) <- glb_txt_vars
if (glb_rsp_var_raw != glb_rsp_var)
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features,
glb_rsp_var_raw)
glbFctrMaxUniqVals <- 23 # default: 20
glb_impute_na_data <- FALSE # or TRUE
glb_mice_complete.seed <- 144 # or any integer
glb_cluster <- FALSE # bid._sp:TRUE # default:FALSE
glb_cluster.seed <- 189 # or any integer
### !_sp
glb_cluster_entropy_var <- NULL # c(glb_rsp_var, as.factor(cut(glb_rsp_var, 3)), default: NULL)
### bid._sp
# glb_cluster_entropy_var <- "sold" #"startprice.log10.cut.fctr"
# glb_exclude_cluster_vars_as_features <- TRUE # default FALSE
### bid._sp
glb_interaction_only_feats_lst <- list()
### bid._sp
glb_interaction_only_feats_lst[["carrier.fctr"]] <- "cellular.fctr"
glb_nzv_freqCut <- 19 # 19 is caret default
glb_nzv_uniqueCut <- 10 # 4 : bid._sp # 10 : caret::default
### bid._sp
# outliers identified by car::outlierTest
glb_obsfit_outliers <- NULL
### bid._sp
# c(NULL # default: NULL
# biddable == 0 & 1; err.abs.fit.sum=423.55172
# # outliers
# , 10813 # next 665 w/ rstudent=-5.091080; biddable=3.263257; err.abs.fit.sum=418.598755
# , 10666 # next 1727 w/ rstudent=-5.163517; biddable=4.293465; err.abs.fit.sum=414.093609
# , 11736 # next 780 w/ rstudent=-5.181343; biddable=5.670483; err.abs.fit.sum=401.817992
# # old biddable importance above this
# , 10781 # next 1323 w/ rstudent=-5.151062; biddable=13.30602; err.abs.fit.sum=396.393721
# #, 10091 # next 91 w/ rstudent=-4.444452; biddable=; err.abs.fit.sum=402.673715 (up)
# #, 10166 # next 560 w/ rstudent=-5.006795; biddable=; err.abs.fit.sum=401.759324 (up)
# #, 10281 # next 281 w/ rstudent=-4.245087; biddable=; err.abs.fit.sum=401.316926 (up)
# #, 10285 # next 285 w/ rstudent=-4.483190; biddable=; err.abs.fit.sum=402.608936 (up)
# #, 10446 # next 445 w/ rstudent=-4.663418; biddable=; err.abs.fit.sum=403.074523 (up)
# #, 10542 # next 1323 w/ rstudent=-5.214517; biddable=; err.abs.fit.sum=401.04205 (up)
# #, 10543 # next 1323 w/ rstudent=-5.214517; biddable=; err.abs.fit.sum=401.04205 (up)
# #, 10561 # next 542 w/ rstudent=-4.736154; biddable=; err.abs.fit.sum=401.56198 (up)
# #, 10631 # next 166 w/ rstudent=-5.073048; biddable=; err.abs.fit.sum=401.556788 (up)
# #, 11330 # next 630 w/ rstudent=-5.117659; biddable=; err.abs.fit.sum=401.732597 (up)
# , 10091, 10166, 10281, 10285, 10446, 10542, 10543, 10561, 10631, 11330
# # biddable=18.93923; err.abs.fit.sum=359.388769
# , 10330 #biddable=19.06084; err.abs.fit.sum=355.895702
# , 10402 #biddable= 0.0 ; err.abs.fit.sum=351.315181
# , 10438 #biddable= 0.0 ; err.abs.fit.sum=347.821527
# , 10624 #biddable= 0.0 ; err.abs.fit.sum=343.724904
# , 10659 #biddable= 0.0 ; err.abs.fit.sum=331.873603
# , 11323 #biddable=10.45901; err.abs.fit.sum=324.929562
# , 11422 #biddable= 0.0 ; err.abs.fit.sum=334.839805 (up)
# biddable == 0; err.abs.fit.sum=26.713317
# , 11448 # outliers; next is 858 w/ rstudent=-5.855132; err.abs.fit.sum=24.212800
# , 11583 # outliers; next is 856 w/ rstudent=-4.792849; err.abs.fit.sum=22.164035
# , 11581 # outliers; next is 743 w/ rstudent=-4.005054; err.abs.fit.sum=18.842901
# , 10837 # outliers; next is 336 w/ rstudent=-5.279215; err.abs.fit.sum=18.124560
# , 11442 # outliers; next is 904 w/ rstudent=-4.474844; err.abs.fit.sum=15.533211
# , 11697 # outliers; next is 874 w/ rstudent=-3.678664; err.abs.fit.sum=13.829375
# , 10799 # .hatvalues == 1; total 8; iPadmini#1; err.abs.fit.sum=13.807283
# , 10017 # .hatvalues == 1; total 7; iPad3#1; err.abs.fit.sum=14.620782 (up)
# , 10027, 10859 # .hatvalues == 1; total 7; iPad1#1; err.abs.fit.sum=14.570246 (up)
# , 10332 # .hatvalues == 1; total 7; iPad4#1; err.abs.fit.sum=13.706467
# , 11759 # .hatvalues == 1; total 6; iPadAir2#1; err.abs.fit.sum=13.643043
# , 10675 # .hatvalues == 1; total 5; iPadAir#1; err.abs.fit.sum=13.623787
# , 11119 # .hatvalues == 1; total 4; iPadmini3#1; err.abs.fit.sum=NA
# , 10017, 10027, 10859 # .hatvalues == 1; total 1; iPad3#1 & iPad1#1; err.abs.fit.sum=13.438903
# biddable == 1; err.abs.fit.sum=361.78243
# , 10813 # outliers; next is 665 w/ rstudent=-5.021180; err.abs.fit.sum=356.83424
# , 10666 # outliers; next is 808 w/ rstudent=-4.764126; err.abs.fit.sum=352.46437
# , 11736 # outliers; next is 665 w/ rstudent=-4.614022; err.abs.fit.sum=348.59977
# , 10542 # outliers; next is 665 w/ rstudent=-4.654923; err.abs.fit.sum=344.18546
# , 11330 # outliers; next is 327 w/ rstudent=-4.628972; err.abs.fit.sum=336.12636
# , 10561 # outliers; next is 56 w/ rstudent=-4.612970; err.abs.fit.sum=329.50309
# , 10166 # outliers; next is 318 w/ rstudent=-4.717238; err.abs.fit.sum=318.50562
# , 10543 # outliers; next is 464 w/ rstudent=-4.811116; err.abs.fit.sum=314.32801
# , 10285 # outliers; next is 21 w/ rstudent=-4.850822; err.abs.fit.sum=310.19008
# #, 10091 # outliers; next is 464 w/ rstudent=-4.941448; err.abs.fit.sum=312.94069 (up)
# #, 10781 # outliers; next is 250 w/ rstudent=-4.793502; err.abs.fit.sum=313.03867 (up)
# , 10446 # outliers; next is 371 w/ rstudent=-4.787578; err.abs.fit.sum=307.15681
# , 10631 # outliers; next is 165 w/ rstudent=-4.130356; err.abs.fit.sum=303.34549
# #, 10330 # outliers; next is 217 w/ rstudent=-4.067684; err.abs.fit.sum=312.75121 (up)
# #, 10402 # outliers; next is 388 w/ rstudent=-4.067684; err.abs.fit.sum=311.84516 (up)
# #, 10659 # outliers; next is 128 w/ rstudent=-3.982911; err.abs.fit.sum=311.84516 (up)
# , 10091, 10781, 10330, 10402, 10659#, 10281 outliers; err.abs.fit.sum=282.381827; iPad4#0=13.806011; iPad4#1=7.799398
# #, 10281 # outliers; next is NA w/ rstudent=NA; err.abs.fit.sum=287.147331 (up); iPad4#0=14.372770; iPad4#1=4.591408
# #, 10624 # outliers; ignored along with 10281 err.abs.fit.sum=289.116467 (up); iPad4#0=; iPad4#1=
# #, 10624 # outliers; ignored w/o 10281 err.abs.fit.sum=286.415040 (up); iPad4#0=; iPad4#1=
# #, 10636 # hatvalues==1; next is 11652; err.abs.fit.sum=290.50254 (up)
# , 11652 # hatvalues==1; next is 10636; err.abs.fit.sum=282.183867
# #err.abs.fit.sum=282.227249
# )
### bid._sp
# influence.measures: car::outlier; rstudent; dffits; hatvalues; dfbeta; dfbetas
#print(outliers <- car::outlierTest(glb_models_lst[["RFE.X.glm"]]$finalModel))
#print(outliers_df <- data.frame(.Bonf.p=outliers$bonf.p))
#model_diags_df <- cbind(glb_fitobs_df, data.frame(.rstudent=stats::rstudent(glb_models_lst[["RFE.X.glm"]]$finalModel)), data.frame(.dffits=stats::dffits(glb_models_lst[["RFE.X.glm"]]$finalModel)), data.frame(.hatvalues=stats::hatvalues(glb_models_lst[["RFE.X.glm"]]$finalModel)))
#print(subset(model_diags_df, is.na(.dffits)))
#print(subset(model_diags_df, .hatvalues == 1))
#dffits_df <- merge(dffits_df, outliers_df, by="row.names", all.x=TRUE); row.names(dffits_df) <- dffits_df$Row.names; dffits_df <- subset(dffits_df, select=-Row.names)
#dffits_df <- merge(dffits_df, glb_fitobs_df, by="row.names", all.x=TRUE); row.names(dffits_df) <- dffits_df$Row.names; dffits_df <- subset(dffits_df, select=-Row.names)
#subset(dffits_df, !is.na(.Bonf.p))
#indep_vars <- myextract_actual_feats(row.names(orderBy(reformulate(c("-", paste0(glb_sel_mdl_id, ".importance"))), glb_featsimp_df))); indep_vars <- indep_vars[!grepl(".fctr", indep_vars, fixed=TRUE)]
#myplot_parcoord(obs_df=model_diags_df[, c(glb_id_var, glb_category_var, ".rstudent", ".dffits", ".hatvalues", glb_rsp_var, "startprice.log10.predict.RFE.X.glmnet", indep_vars[1:20])], obs_ix=row.names(model_diags_df) %in% names(outliers$rstudent)[1], id_var=glb_id_var, category_var=glb_category_var)
#model_diags_df[row.names(model_diags_df) %in% names(outliers$rstudent)[c(1:2)], ]
#ctgry_diags_df <- model_diags_df[model_diags_df[, glb_category_var] %in% c("Unknown#0"), ]
#myplot_parcoord(obs_df=ctgry_diags_df[, c(glb_id_var, glb_category_var, ".rstudent", ".dffits", ".hatvalues", glb_rsp_var, "startprice.log10.predict.RFE.X.glmnet", indep_vars[1:20])], obs_ix=row.names(ctgry_diags_df) %in% names(outliers$rstudent)[1], id_var=glb_id_var, category_var=glb_category_var)
#table(glb_fitobs_df[model_diags_df[, glb_category_var] %in% c("iPad1#1"), "startprice.log10.cut.fctr"])
#glb_fitobs_df[model_diags_df[, glb_category_var] %in% c("iPad1#1"), c(glb_id_var, "startprice")]
# No outliers & .dffits == NaN
#myplot_parcoord(obs_df=model_diags_df[, c(glb_id_var, glb_category_var, glb_rsp_var, "startprice.log10.predict.RFE.X.glmnet", indep_vars[1:10])], obs_ix=seq(1:nrow(model_diags_df))[is.na(model_diags_df$.dffits)], id_var=glb_id_var, category_var=glb_category_var)
#dffits_ctgry_df <- subset(dffits_df, prdl.descr.my.fctr %in% c("Unknown#0"))
#myplot_parcoord(obs_df=dffits_ctgry_df[, c(glb_id_var, glb_category_var, ".dffits", ".Bonf.p", glb_rsp_var, "startprice.log10.predict.RFE.X.glmnet", indep_vars[1:5])], obs_ix=seq(1:nrow(dffits_ctgry_df))[!is.na(dffits_ctgry_df$.Bonf.p)], id_var=glb_id_var, category_var=glb_category_var)
#
#car::influenceIndexPlot(glb_models_lst[["RFE.X.glm"]]$finalModel, id.n=3)
myplot_parcoord <- function (obs_df, obs_ix=1:nrow(obs_df), id_var=".rownames", category_var=NULL) {
# Setup id_df & remove id_var from range computation
if (id_var != ".rownames") {
id_df <- obs_df[obs_ix, id_var, FALSE]
obs_df <- obs_df[, setdiff(names(obs_df), id_var), FALSE]
} else id_df <- data.frame(.rownames=row.names(obs_df)[obs_ix])
# Setup category_var -> Create a facet ???
category_df <- id_df
if (is.null(category_var)) {
category_var <- ".category"; category_df[, category_var] <- as.factor(0)
} else {
category_df[, category_var] <- obs_df[obs_ix, category_var]
obs_df <- obs_df[, setdiff(names(obs_df), category_var), FALSE]
}
ranges_mtrx <- apply(obs_df, 2L, range, na.rm = TRUE)
obs_scld_df <- as.data.frame(apply(obs_df, 2L,
function(feat) { feat_rng <- max(feat, na.rm = TRUE) - min(feat, na.rm = TRUE);
feat_rng <- ifelse(feat_rng == 0, 1, feat_rng);
return((feat - min(feat, na.rm = TRUE)) / feat_rng) }))
obsT_df <- as.data.frame(t(obs_df))
names(obsT_df) <- paste(".obs", names(obsT_df), sep=".");
obsT_df$.var.name <- row.names(obsT_df)
obsT_df$.var.pos <- 1:length(row.names(obsT_df))
obsST_df <- as.data.frame(t(obs_scld_df))
names(obsST_df) <- paste(".obs", names(obsST_df), sep=".");
obsST_df$.var.name <- row.names(obsST_df)
obsST_df$.var.pos <- 1:length(row.names(obsST_df))
plt_violin_df <- tidyr::gather(obsST_df, key=obs, value=value, -.var.name, -.var.pos)
obsHST_df <- as.data.frame(t(obs_scld_df[obs_ix, ]));
names(obsHST_df) <- as.character(id_df[, id_var])
obsHST_df$.var.name <- row.names(obsHST_df)
obsHST_df$.var.pos <- 1:length(row.names(obsHST_df))
#plt_df <- tidyr::gather(xt_df, key=obs, value=value, -c(.var.name, .var.pos))
# plt_df <- tidyr::gather_(xt_df, key=interp(id_var), value="value", quote(-c(.var.name, .var.pos)))
plt_obsHST_df <- tidyr::gather_(obsHST_df, key=interp(id_var), value="value",
-grep("(\\.var\\.name|\\.var\\.pos)", names(obsHST_df)))
ranges_df <- cbind(as.data.frame(ranges_mtrx), data.frame(.type=c("min", "max")))
ranges_df <- tidyr::gather(ranges_df, key=.var, value=value, -.type)
ranges_df$.y <- ifelse(ranges_df$.type == "min", -0.05, 1.05)
ranges_df <- merge(ranges_df, obsT_df[, c(".var.name", ".var.pos")],
by.x=".var", by.y=".var.name", all.x=TRUE)
ranges_df$.x <- ranges_df$.var.pos
ranges_df <- subset(ranges_df, select=-.var.pos)
plt_obsHST_df <- merge(plt_obsHST_df, category_df, x.all=TRUE)
# plt_obsHST_df[, category_var] <- NA
# plt_obsHST_df[plt_obsHST_df[, id_var] == 11448, glb_category_var] <- "Unknown#0"
# plt_obsHST_df[plt_obsHST_df[, id_var] == 11581, glb_category_var] <- "iPad4#1"
# plt_obsHST_df[plt_obsHST_df[, id_var] == 11583, glb_category_var] <- "Unknown#0"
gp <- ggplot(plt_obsHST_df, aes(x=reorder(.var.name, .var.pos), y=value)) +
geom_violin(data=plt_violin_df, aes(x=reorder(.var.name, .var.pos), y=value),
color="grey80", scale="width") +
geom_line(data=plt_obsHST_df,
aes_string(group=id_var, color=id_var, linetype=category_var), size=1) +
geom_point(data=plt_obsHST_df, aes_string(shape=category_var), size=3) +
scale_color_brewer(type="qual", palette="Set1") +
geom_vline(xintercept=1:length(names(obs_df)), color="grey50") +
geom_text(data=ranges_df,
aes_string(x=".x", y=".y", label="myformat_number(value)"),
size=3.5) +
theme(axis.text.x=element_text(hjust=1, angle=45),
axis.text.y=element_blank(), axis.ticks.y=element_blank()) +
xlab("") + ylab("")
# ggtitle("Dummy")
return(gp)
}
# myplot_parcoord(obs_df=glb_fitobs_df[, c(glb_id_var, glb_rsp_var,
# "startprice.log10.predict.RFE.X.glmnet",
# indep_vars[1:5])], obs_ix=hatobs_ix, id_var=glb_id_var)
# myplot_parcoord(x=glb_fitobs_df[, c(glb_rsp_var, "startprice.log10.predict.RFE.X.glmnet",
# indep_vars[1:2])], obs_ix=hatobs_ix)
# hatvals <- hatvalues(glb_models_lst[["RFE.X.glm"]]$finalModel)
# hatobs_ix <- which(hatvals == max(hatvals))
# MASS::parcoord(x=glb_fitobs_df[, c(glb_rsp_var, "startprice.log10.predict.RFE.X.glmnet",
# indep_vars[1:2])], var.label=TRUE)
#plot(hatvalues(glb_models_lst[["RFE.X.glm"]]$finalModel), type = "h")
#glb_fitobs_df[which(row.names(glb_fitobs_df) %in% c("972")), c(glb_id_var, glb_rsp_var, glb_rsp_var_raw, "sold", glb_category_var)]
#all.equal(glb_models_lst[[glb_sel_mdl_id]], glb_models_lst[[glb_fin_mdl_id]])
glb_obstrn_outliers <- c(glb_obsfit_outliers
)
#car::outlierTest(glb_models_lst[["RFE.X.glm"]]$finalModel)
#glb_trnobs_df[which(row.names(glb_fitobs_df) %in% c("972")), c(glb_id_var, glb_rsp_var, glb_rsp_var_raw, "sold", glb_category_var)]
glb_rfe_fit_sizes <- NULL
### bid0_sp
# c(106, 111, 116, 120, 128) # or NULL c(8, 16, 32, 64, 128, 140)
### bid1_sp
# c(8, 11, 16, 21, 32, 64, 128)
glb_models_lst <- list(); glb_models_df <- data.frame()
# Regression
if (glb_is_regression) {
glb_mdl_methods <- c(NULL
# deterministic
#, "lm",
, "glm"
#, "bayesglm" # crashing w/ parallel processing
, "glmnet", "rpart"
# non-deterministic
, "gbm", "rf"
# Unknown
#, "nnet" , "avNNet" # predicts 1 for all obs in bid0_sp # runs 25 models per cv sample for tunelength=5
, "svmLinear", "svmLinear2"
#, "svmPoly" # crashing w/ parallel processing #, "svmPoly" runs 75 models per cv sample for tunelength=5
#, "svmRadial" # crashing w/ parallel processing
, "earth", "bagEarth" # Takes a long time
#, "parRF" # crashing w/ parallel processing
)
} else
# Classification - Add ada,bagEarth (auto feature selection)
if (glb_is_binomial)
glb_mdl_methods <- c(NULL
# deterministic
, "glm"
, "bayesglm" # crashing w/ parallel processing
, "glmnet", "rpart"
# non-deterministic
, "gbm", "rf"
) else
glb_mdl_methods <- c("rpart", "rf", "gbm")
glb_mdl_family_lst <- list(); glb_mdl_feats_lst <- list()
# family: Choose from c("RFE.X", "CSM.X", "All.X", "Best.Interact")
# methods: Choose from c(NULL, <method>, glb_mdl_methods)
glb_mdl_family_lst[["RFE.X"]] <- glb_mdl_methods # non-NULL list is mandatory
glb_mdl_family_lst[["All.X"]] <- "glmnet" # non-NULL list is mandatory
glb_mdl_family_lst[["Best.Interact"]] <- "glmnet" # non-NULL list is mandatory
### bid1_sp
# glb_mdl_family_lst[["CSM.X"]] <- "glmnet"
# glb_mdl_feats_lst[["CSM.X"]] <- c(NULL
# # from RFE.X
# , "startprice.dgt1.is9", "startprice.dcm2.is9", "startprice.dcm1.is9", "startprice.dgt2.is9"
# #, "condition.fctr"
# , "prdl.descr.my.fctr", "color.fctr"
# #, "D.ratio.weight.sum.wrds.n"
# , "cellular.fctr", "cellular.fctr:carrier.fctr"
#
# # from RFE.X.Interact
# , "cellular.fctr:prdl.descr.my.fctr", "cellular.fctr:startprice.dgt2.is9", "cellular.fctr:startprice.dgt1.is9", "cellular.fctr:color.fctr"
# , "cellular.fctr:condition.fctr" # RMSE up with keeping condition.fctr in the model
# # RMSE & R.sq up with removing condition.fctr from the model
# , "cellular.fctr:D.ratio.weight.sum.wrds.n"
# )
### bid1_sp
### !_sp
# glb_mdl_family_lst[["CSM.X"]] <- "glmnet"
# glb_mdl_feats_lst[["CSM.X"]] <- c(NULL
# , "D.ratio.sum.TfIdf.wrds.n", "D.TfIdf.sum.stem.stop.Ratio",
# , "D.npnct15.log", "D.npnct03.log", "D.wrds.n.log", "D.chrs.n.log")
# indep_vars <- union(setdiff(indep_vars, interact_vars_vctr),
# paste(glb_category_var, interact_vars_vctr,
# sep=ifelse(grepl("\\.fctr", glb_category_var), "*", ".fctr*")))
# indep_vars <- union(setdiff(indep_vars,
# c("startprice.log.diff", "startprice.unit9", "biddable", "cellular.fctr", "carrier.fctr")),
# c("startprice.log.diff*biddable", "startprice.unit9*biddable", "cellular.fctr*carrier.fctr"))
### !_sp
# Check if interaction features make fit better
# Check if tuning parameters make fit better
glb_tune_models_df <- data.frame()
#RFE.X.avNNet
### bid0_sp
# size=[1] 3 5 7 9; decay=[0] 1e-04 0.001 0.01 0.1; bag=[FALSE]; RMSE=1.3300906
### bid1_sp
# size=[1] 3 5 7 9; decay=0 0.0001 [0.001] 0.01 0.1; bag=[FALSE]; RMSE=0.9285472
### bid0&1_sp
#RFE.X.bagEarth
### bid0_sp
#RFE.X.bagEarth degree=[1]; nprune=[33]; RMSE=0.1507259
### bid1_sp
#RFE.X.bagEarth degree=[1]; nprune=[32]; RMSE=0.6379639
#RFE.X.bagEarth degree=[1] 2 3; nprune=8 16 32 64 [128]; RMSE=0.6334405
#RFE.X.bagEarth degree=1 [2]; nprune=16 32 64 128 [256]; RMSE=0.6211924
#RFE.X.bagEarth degree=1 [2]; nprune=64 128 200 225 [256]; RMSE=0.6320776 (up)
#RFE.X.bagEarth degree=[1] 2; nprune=64 128 225 256 [275]; RMSE=0.640644 (up)
#RFE.X.bagEarth degree=1 [2] 3; nprune=64 128 200 [256] 300; RMSE=0.6496039 (up)
#RFE.X.bagEarth degree=1 [2] 3; nprune=32 64 128 256 [512]; RMSE=0.6404529 (up)
#RFE.X.bagEarth degree=1 [2] 3; nprune=64 128 256 512 [1024]; RMSE=0.6486663 (up)
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "bagEarth", parameter = "nprune", vals = "256")
# ,data.frame(method = "bagEarth", parameter = "degree", vals = "2")
# ))
### bid0&1_sp
### bid0_sp
#RFE.X.earth degree=[1]; nprune=2 [9] 17 25 33; RMSE=0.1334478
### bid0_sp
#RFE.X.gbm
### bid0_sp
# shrinkage=[0.1]; n.trees=50 100 150 [200] 250; RMSE=0.2062651
# shrinkage=0.00 0.05 0.10 0.15 [0.20]; n.trees=50 [100] 150 200 250; interaction.depth=1 [2] 3 4 5; n.minobsinnode=[10]; RMSE=0.2019453
# shrinkage=0.00 0.05 [0.10] 0.15 0.20; n.trees=50 100 150 200 [250]; interaction.depth=[1] 2 3 4 5; n.minobsinnode=[10]; RMSE=0.2008313
# shrinkage=0.05 [0.10] 0.15 0.20 0.25; n.trees=100 150 200 [250] 300; interaction.depth=[1] 2 3 4 5; n.minobsinnode=[10]; RMSE=0.2008313
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "gbm", parameter = "shrinkage", min = 0.05, max = 0.25, by = 0.05)
# ,data.frame(method = "gbm", parameter = "n.trees", min = 100, max = 300, by = 50)
# ,data.frame(method = "gbm", parameter = "interaction.depth", min = 1, max = 5, by = 1)
# ,data.frame(method = "gbm", parameter = "n.minobsinnode", min = 10, max = 10, by = 10)
# #seq(from=0.05, to=0.25, by=0.05)
# ))
### bid1_sp
# shrinkage=[0.1]; n.trees=50 100 150 200 [250]; interaction.depth=1 2 3 4 [5]; n.minobsinnode=[10]; RMSE=0.5054172
# shrinkage=0.03 [0.04] 0.05 0.06 0.07; n.trees=100 [150] 200 250 300; interaction.depth=2 3 4 5 [6]; n.minobsinnode=6 [8] 10 12 14; RMSE=0.5036430
# shrinkage=0.03 [0.04] 0.05 0.06 0.07; n.trees=100 150 [200] 250 300; interaction.depth=3 4 5 [6] 7; n.minobsinnode=6 8 [10] 12 14; RMSE=0.502774
# shrinkage=0.04; n.trees=200; interaction.depth=6; n.minobsinnode=10; RMSE=0.502774
# shrinkage=[0.05] 0.10 0.15 0.20 0.25; n.trees=100 [150] 200 250 300; interaction.depth=2 3 [4] 5 6; n.minobsinnode=[10]; RMSE=0.5058678 (up)
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "gbm", parameter = "shrinkage", vals = "0.04")
# ,data.frame(method = "gbm", parameter = "n.trees", vals = "200")
# ,data.frame(method = "gbm", parameter = "interaction.depth", vals = "6")
# ,data.frame(method = "gbm", parameter = "n.minobsinnode", vals = "10")
# ))
### bid0&1_sp
#RFE.X.glmnet
### bid1_sp
# alpha=0.100 [0.325] 0.550 0.775 1.000; lambda=0.0005232693 0.0024288010 0.0112734954 [0.0523269304] 0.2428800957; RMSE=0.6164891
### bid1_sp
#RFE.X.nnet
### bid0_sp
# size=[1] 3 5 7 9; decay=[0] 1e-04 0.001 0.01 0.1; RMSE=1.3300906
### bid1_sp
# size=1 3 5 7 [9]; decay=0e+00 1e-04 1e-03 1e-02 [1e-01]; RMSE=0.9289109
# size=3 5 [7] 9 11; decay=0.0001 0.001 0.01 [0.1] 0.2; RMSE=0.9287422
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "nnet", parameter = "size", vals = "3 5 7 9 11")
# ,data.frame(method = "nnet", parameter = "decay", vals = "0.0001 0.0010 0.0100 0.1000 0.2000")
# ))
### bid0&1_sp
#RFE.X.rf
### bid0_sp
# mtry=2 35 [68] 101 134; RMSE=0.1331992
# mtry=2 35 68 [101] 134; RMSE=0.1339974
### bid0_sp
#RFE.X.rpart
### bid0_sp
# cp=[0.03230142] 0.06012801 0.09395662 0.12251081 0.35258370; RMSE=0.1771138
# cp=0.020 [0.025] 0.030 0.035 0.040; RMSE=0.1770237
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method="rpart", parameter="cp", min=0.02, max=0.04, by=0.005)
# ))
### bid1_sp
# cp=[0.008081388] 0.016191995 0.027590245 0.299848193 0.361621486; RMSE=0.5294398
# cp=[0.005] 0.006 0.007 0.008 0.009 0.010; RMSE=0.522678
# cp=0.001 [0.003] 0.005 0.007 0.009; RMSE=0.5186586
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method="rpart", parameter="cp", min=0.001, max=0.010, by=0.002)
# ))
### bid0&1_sp
#RFE.X.svmLinear
### bid0_sp
# C=[1]; RMSE=0.1374094
# C=1e-02 [0.1] 5e-01 1e+00 2e+00 3e+00 4e+00 1e+01 1e+02; RMSE=0.1271318
# C=0.01 0.05 [0.10] 0.50 1.00 2.00 3.00 4.00; RMSE=0.1271318; 0.1296718
### bid1_sp
# C=[1]; RMSE=0.6614060
# C=1e-02 [1e-01] 1e+00 1e+01 1e+02; RMSE=0.6373977
# C=[0.05] 0.10 0.50 1.00 10.00; RMSE=0.6324697
# C=0.01 [0.05] 0.10 0.50 1.00; RMSE=0.6324697
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "svmLinear", parameter = "C", vals = "0.01 0.05 0.1 0.5 1")
# ))
### bid0&1_sp
#RFE.X.svmLinear2
### bid0_sp
# cost=[0.25] 0.50 1.00 2.00 4.00; RMSE=0.1276354
# cost=0.0625 0.1250 [0.25] 0.50 1.00; RMSE=0.1276354
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "svmLinear2", parameter = "cost", vals = "0.0625 0.125 0.25 0.5 1")
### bid1_sp
# cost=[0.25] 0.50 1.00 2.00 4.00; RMSE=0.6483622
# cost=[0.0625] 0.1250 0.25 0.50 1.00; RMSE=0.6335311
# cost=0.0312 [0.0625] 0.1250 0.25 0.50; RMSE=0.6335311
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method = "svmLinear2", parameter = "cost", vals = "0.0312 0.0625 0.125 0.25 0.50")
# ))
### bid0&1_sp
#RFE.X.svmPoly
### bid0_sp
# degree=[1] 2 3; scale=0.001 0.01 [0.1] 1 10; C=0.25 0.50 1.00 [2.00] 4.00; RMSE=0.1276130
# degree=[1] 2 3 4 5; scale=0.01 0.05 [0.1] 0.5 1; C=0.50 1.00 [2.00] 3.00 4.00; RMSE=0.1276130
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method="svmPoly", parameter="degree", min=1, max=5, by=1) #seq(1, 5, 1)
# ))
# glb_tune_models_df <- myrbind_df(glb_tune_models_df, rbind(data.frame()
# ,data.frame(method="svmPoly", parameter="scale", vals="0.01, 0.05, 0.1, 0.5, 1")
# ,data.frame(method="svmPoly", parameter="C", vals="0.50, 1.00, 2.00, 3.00, 4.00")
# ))
### bid0_sp
#RFE.X.svmRadial
### bid0_sp
# sigma=[0.08674323]; C=0.25 0.50 1.00 [2.00] 4.00; RMSE=0.1614957
### bid0_sp
#data.frame(parameter="mtry", min=080, max=100, by=10),
#glb_to_sav(); all.equal(sav_models_df, glb_models_df)
#glb_models_df <- subset(sav_models_df, id != "RFE.X.gbm"); print(sort(glb_models_df$id))
glb_preproc_methods <- NULL
### bid0_sp
# c("YeoJohnson", "center.scale",
# # crashes with train: all the RMSE metric values are missing
# # probably due to interaction vars
# "range", "pca", "ica",
# "spatialSign")
### bid0_sp
### bid1_sp
# c("YeoJohnson", "center.scale", "range", "pca", "ica", "spatialSign")
### bid1_sp
# Baseline prediction model feature(s)
glb_Baseline_mdl_var <- NULL # or c("<col_name>")
glb_model_metric_terms <- NULL # or matrix(c(
# 0,1,2,3,4,
# 2,0,1,2,3,
# 4,2,0,1,2,
# 6,4,2,0,1,
# 8,6,4,2,0
# ), byrow=TRUE, nrow=5)
glb_model_metric <- NULL # or "<metric_name>"
glb_model_metric_maximize <- NULL # or FALSE (TRUE is not the default for both classification & regression)
glb_model_metric_smmry <- NULL # or function(data, lev=NULL, model=NULL) {
# confusion_mtrx <- t(as.matrix(confusionMatrix(data$pred, data$obs)))
# #print(confusion_mtrx)
# #print(confusion_mtrx * glb_model_metric_terms)
# metric <- sum(confusion_mtrx * glb_model_metric_terms) / nrow(data)
# names(metric) <- glb_model_metric
# return(metric)
# }
glb_rcv_n_folds <- 3 # or NULL
glb_rcv_n_repeats <- 3 # or NULL
glb_clf_proba_threshold <- NULL # 0.5
# Model selection criteria
if (glb_is_regression)
glb_model_evl_criteria <- c("min.RMSE.OOB", "max.R.sq.OOB", "max.Adj.R.sq.fit")
#glb_model_evl_criteria <- c("min.RMSE.fit", "max.R.sq.fit", "max.Adj.R.sq.fit")
if (glb_is_classification) {
if (glb_is_binomial)
glb_model_evl_criteria <-
c("max.auc.OOB", "max.Accuracy.OOB", "max.Kappa.OOB", "min.aic.fit") else
glb_model_evl_criteria <- c("max.Accuracy.OOB", "max.Kappa.OOB")
}
# select from NULL [no ensemble models], "auto" [all models better than MFO or Baseline], c(mdl_ids in glb_models_lst) [Typically top-rated models in auto]
glb_mdl_ensemble <- NULL
### bid0_sp
# c("RFE.X.glm"
# #, "RFE.X.bayesglm"
# , "RFE.X.glmnet", "RFE.X.rpart", "RFE.X.gbm", "RFE.X.rf", "RFE.X.svmLinear", "RFE.X.svmLinear2"
# #, "RFE.X.svmPoly", "RFE.X.svmRadial"
# , "RFE.X.earth", "RFE.X.bagEarth", "RFE.X.Interact.glmnet", "RFE.X.YeoJohnson.glmnet", "RFE.X.center.scale.glmnet", "RFE.X.spatialSign.glmnet")
### bid1_sp
# "auto"; err.abs.fit.sum=76.699774; min.RMSE.fit=0.2186429
# "RFE.X.*"; err.abs.fit.sum=; min.RMSE.fit=0.221114
# c("RFE.X.spatialSign.rf", "RFE.X.YeoJohnson.rf", "RFE.X.center.scale.rf", "RFE.X.rf", "RFE.X.avNNet", "RFE.X.bagEarth", "RFE.X.earth", "RFE.X.gbm", "RFE.X.glmnet", "RFE.X.nnet", "RFE.X.svmLinear2", "RFE.X.glm", "RFE.X.svmLinear", "RFE.X.rpart")
### bid1_sp
glb_sel_mdl_id <- "RFE.X.glmnet" # NULL #select from c(NULL, "RFE.X.glmnet")
glb_fin_mdl_id <- NULL #select from c(NULL, glb_sel_mdl_id)
glb_dsp_cols <- c("sold", ".grpid", "color", "condition", "cellular", "carrier", "storage")
glb_out_obs <- NULL # "all" for bid._sp # select from c(NULL, "all", "new", "trn")
glb_out_vars_lst <- list()
# glb_id_var will be the first output column, by default
### !_sp
glb_out_vars_lst[["Probability1"]] <- "%<d-% paste0(glb_rsp_var_out, glb_fin_mdl_id, '.prob')"
### !_sp
### bid._sp
# glb_out_vars_lst[[glb_rsp_var_raw]] <- glb_rsp_var_raw
# glb_out_vars_lst[[paste0(head(unlist(strsplit(glb_rsp_var_out, "")), -1), collapse = "")]] <-
# "%<d-% paste0(glb_rsp_var_out, glb_fin_mdl_id)"
### bid._sp
# Depict process
glb_analytics_pn <- petrinet(name="glb_analytics_pn",
trans_df=data.frame(id=1:6,
name=c("data.training.all","data.new",
"model.selected","model.final",
"data.training.all.prediction","data.new.prediction"),
x=c( -5,-5,-15,-25,-25,-35),
y=c( -5, 5, 0, 0, -5, 5)
),
places_df=data.frame(id=1:4,
name=c("bgn","fit.data.training.all","predict.data.new","end"),
x=c( -0, -20, -30, -40),
y=c( 0, 0, 0, 0),
M0=c( 3, 0, 0, 0)
),
arcs_df=data.frame(
begin=c("bgn","bgn","bgn",
"data.training.all","model.selected","fit.data.training.all",
"fit.data.training.all","model.final",
"data.new","predict.data.new",
"data.training.all.prediction","data.new.prediction"),
end =c("data.training.all","data.new","model.selected",
"fit.data.training.all","fit.data.training.all","model.final",
"data.training.all.prediction","predict.data.new",
"predict.data.new","data.new.prediction",
"end","end")
))
#print(ggplot.petrinet(glb_analytics_pn))
print(ggplot.petrinet(glb_analytics_pn) + coord_flip())## Loading required package: grid
glb_analytics_avl_objs <- NULL
glb_chunks_df <- myadd_chunk(NULL, "import.data")## label step_major step_minor label_minor bgn end elapsed
## 1 import.data 1 0 0 11.272 NA NAStep 1.0: import data
chunk option: eval=
#glb_chunks_df <- myadd_chunk(NULL, "import.data")
glb_to_sav <- function() {
sav_allobs_df <<- glb_allobs_df
sav_trnobs_df <<- glb_trnobs_df
if (any(grepl("glb_fitobs_df", ls(envir=globalenv()), fixed=TRUE)) &&
!is.null(glb_fitobs_df)) sav_fitobs_df <<- glb_fitobs_df
if (any(grepl("glb_OOBobs_df", ls(envir=globalenv()), fixed=TRUE)) &&
!is.null(glb_OOBobs_df)) sav_OOBobs_df <<- glb_OOBobs_df
if (any(grepl("glb_newobs_df", ls(envir=globalenv()), fixed=TRUE)) &&
!is.null(glb_newobs_df)) {
#print("Attempting to save glb_newobs_df...")
sav_newobs_df <<- glb_newobs_df
}
if (any(grepl("glb_ctgry_df", ls(envir=globalenv()), fixed=TRUE)) &&
!is.null(glb_ctgry_df)) sav_ctgry_df <<- glb_ctgry_df
if (!is.null(glb_models_lst )) sav_models_lst <<- glb_models_lst
if (!is.null(glb_models_df )) sav_models_df <<- glb_models_df
if (any(grepl("glb_feats_df", ls(envir=globalenv()), fixed=TRUE)) &&
!is.null(glb_feats_df)) sav_feats_df <<- glb_feats_df
if (any(grepl("glb_featsimp_df", ls(envir=globalenv()), fixed=TRUE)) &&
!is.null(glb_featsimp_df)) sav_featsimp_df <<- glb_featsimp_df
}
glb_trnobs_df <- myimport_data(url=glb_trnng_url, comment="glb_trnobs_df",
force_header=TRUE)## [1] "Reading file ./data/eBayiPadTrain.csv..."
## [1] "dimensions of data in ./data/eBayiPadTrain.csv: 1,861 rows x 11 cols"
## description
## 1 iPad is in 8.5+ out of 10 cosmetic condition!
## 2 Previously used, please read description. May show signs of use such as scratches to the screen and
## 3
## 4
## 5 Please feel free to buy. All products have been thoroughly inspected, cleaned and tested to be 100%
## 6
## biddable startprice condition cellular carrier color
## 1 0 159.99 Used 0 None Black
## 2 1 0.99 Used 1 Verizon Unknown
## 3 0 199.99 Used 0 None White
## 4 0 235.00 New other (see details) 0 None Unknown
## 5 0 199.99 Seller refurbished Unknown Unknown Unknown
## 6 1 175.00 Used 1 AT&T Space Gray
## storage productline sold UniqueID
## 1 16 iPad 2 0 10001
## 2 16 iPad 2 1 10002
## 3 16 iPad 4 1 10003
## 4 16 iPad mini 2 0 10004
## 5 Unknown Unknown 0 10005
## 6 32 iPad mini 2 1 10006
## description
## 65
## 283 Pristine condition, comes with a case and stylus.
## 948 \211\333\317Used Apple Ipad 16 gig 1st generation in Great working condition and 100% functional.Very little
## 1354
## 1366 Item still in complete working order, minor scratches, normal wear and tear but no damage. screen is
## 1840
## biddable startprice condition cellular carrier color
## 65 0 195.00 Used 0 None Unknown
## 283 1 20.00 Used 0 None Unknown
## 948 0 110.00 Seller refurbished 0 None Black
## 1354 0 300.00 Used 0 None White
## 1366 1 125.00 Used Unknown Unknown Unknown
## 1840 0 249.99 Used 1 Sprint Space Gray
## storage productline sold UniqueID
## 65 16 iPad mini 0 10065
## 283 64 iPad 1 0 10283
## 948 32 iPad 1 0 10948
## 1354 16 iPad Air 1 11354
## 1366 Unknown iPad 1 1 11366
## 1840 16 iPad Air 1 11840
## description
## 1856 Overall item is in good condition and is fully operational and ready to use. Comes with box and
## 1857 Used. Tested. Guaranteed to work. Physical condition grade B+ does have some light scratches and
## 1858 This item is brand new and was never used; however, the box and/or packaging has been opened.
## 1859
## 1860 This unit has minor scratches on case and several small scratches on the display. \nIt is in
## 1861 30 Day Warranty. Fully functional engraved iPad 1st Generation with signs of normal wear which
## biddable startprice condition cellular carrier
## 1856 0 89.50 Used 1 AT&T
## 1857 0 239.95 Used 0 None
## 1858 0 329.99 New other (see details) 0 None
## 1859 0 400.00 New 0 None
## 1860 0 89.00 Seller refurbished 0 None
## 1861 0 119.99 Used 1 AT&T
## color storage productline sold UniqueID
## 1856 Unknown 16 iPad 1 0 11856
## 1857 Black 32 iPad 4 1 11857
## 1858 Space Gray 16 iPad Air 0 11858
## 1859 Gold 16 iPad mini 3 0 11859
## 1860 Black 64 iPad 1 1 11860
## 1861 Black 64 iPad 1 0 11861
## 'data.frame': 1861 obs. of 11 variables:
## $ description: chr "iPad is in 8.5+ out of 10 cosmetic condition!" "Previously used, please read description. May show signs of use such as scratches to the screen and " "" "" ...
## $ biddable : int 0 1 0 0 0 1 1 0 1 1 ...
## $ startprice : num 159.99 0.99 199.99 235 199.99 ...
## $ condition : chr "Used" "Used" "Used" "New other (see details)" ...
## $ cellular : chr "0" "1" "0" "0" ...
## $ carrier : chr "None" "Verizon" "None" "None" ...
## $ color : chr "Black" "Unknown" "White" "Unknown" ...
## $ storage : chr "16" "16" "16" "16" ...
## $ productline: chr "iPad 2" "iPad 2" "iPad 4" "iPad mini 2" ...
## $ sold : int 0 1 1 0 0 1 1 0 1 1 ...
## $ UniqueID : int 10001 10002 10003 10004 10005 10006 10007 10008 10009 10010 ...
## - attr(*, "comment")= chr "glb_trnobs_df"
## NULL# glb_trnobs_df <- read.delim("data/hygiene.txt", header=TRUE, fill=TRUE, sep="\t",
# fileEncoding='iso-8859-1')
# glb_trnobs_df <- read.table("data/hygiene.dat.labels", col.names=c("dirty"),
# na.strings="[none]")
# glb_trnobs_df$review <- readLines("data/hygiene.dat", n =-1)
# comment(glb_trnobs_df) <- "glb_trnobs_df"
# glb_trnobs_df <- data.frame()
# for (symbol in c("Boeing", "CocaCola", "GE", "IBM", "ProcterGamble")) {
# sym_trnobs_df <-
# myimport_data(url=gsub("IBM", symbol, glb_trnng_url), comment="glb_trnobs_df",
# force_header=TRUE)
# sym_trnobs_df$Symbol <- symbol
# glb_trnobs_df <- myrbind_df(glb_trnobs_df, sym_trnobs_df)
# }
# glb_trnobs_df <-
# glb_trnobs_df %>% dplyr::filter(Year >= 1999)
if (glb_is_separate_newobs_dataset) {
glb_newobs_df <- myimport_data(url=glb_newdt_url, comment="glb_newobs_df",
force_header=TRUE)
# To make plots / stats / checks easier in chunk:inspectORexplore.data
glb_allobs_df <- myrbind_df(glb_trnobs_df, glb_newobs_df);
comment(glb_allobs_df) <- "glb_allobs_df"
} else {
glb_allobs_df <- glb_trnobs_df; comment(glb_allobs_df) <- "glb_allobs_df"
if (!glb_split_entity_newobs_datasets) {
stop("Not implemented yet")
glb_newobs_df <- glb_trnobs_df[sample(1:nrow(glb_trnobs_df),
max(2, nrow(glb_trnobs_df) / 1000)),]
} else if (glb_split_newdata_method == "condition") {
glb_newobs_df <- do.call("subset",
list(glb_trnobs_df, parse(text=glb_split_newdata_condition)))
glb_trnobs_df <- do.call("subset",
list(glb_trnobs_df, parse(text=paste0("!(",
glb_split_newdata_condition,
")"))))
} else if (glb_split_newdata_method == "sample") {
require(caTools)
set.seed(glb_split_sample.seed)
split <- sample.split(glb_trnobs_df[, glb_rsp_var_raw],
SplitRatio=(1-glb_split_newdata_size_ratio))
glb_newobs_df <- glb_trnobs_df[!split, ]
glb_trnobs_df <- glb_trnobs_df[split ,]
} else if (glb_split_newdata_method == "copy") {
glb_trnobs_df <- glb_allobs_df
comment(glb_trnobs_df) <- "glb_trnobs_df"
glb_newobs_df <- glb_allobs_df
comment(glb_newobs_df) <- "glb_newobs_df"
} else stop("glb_split_newdata_method should be %in% c('condition', 'sample', 'copy')")
comment(glb_newobs_df) <- "glb_newobs_df"
myprint_df(glb_newobs_df)
str(glb_newobs_df)
if (glb_split_entity_newobs_datasets) {
myprint_df(glb_trnobs_df)
str(glb_trnobs_df)
}
} ## [1] "Reading file ./data/eBayiPadTest.csv..."
## [1] "dimensions of data in ./data/eBayiPadTest.csv: 798 rows x 10 cols"
## description
## 1 like new
## 2 Item is in great shape. I upgraded to the iPad Air 2 and don't need the mini any longer, even though
## 3 This iPad is working and is tested 100%. It runs great. It is in good condition. Cracked digitizer.
## 4
## 5 Grade A condition means that the Ipad is 100% working condition. Cosmetically 8/9 out of 10 - Will
## 6 Brand new factory sealed iPad in an OPEN BOX...THE BOX ITSELF IS HEAVILY DISTRESSED(see
## biddable startprice condition cellular carrier color
## 1 0 105.00 Used 1 AT&T Unknown
## 2 0 195.00 Used 0 None Unknown
## 3 0 219.99 Used 0 None Unknown
## 4 1 100.00 Used 0 None Unknown
## 5 0 210.99 Manufacturer refurbished 0 None Black
## 6 0 514.95 New other (see details) 0 None Gold
## storage productline UniqueID
## 1 32 iPad 1 11862
## 2 16 iPad mini 2 11863
## 3 64 iPad 3 11864
## 4 16 iPad mini 11865
## 5 32 iPad 3 11866
## 6 64 iPad Air 2 11867
## description
## 1 like new
## 142 iPad mini 1st gen wi-fi 16gb is in perfect working order.
## 309 In excellent condition. Minor scratches on the back. Screen in mint condition. Comes in original
## 312 iPad is in Great condition, the screen is in great condition showing only a few minor scratches, the
## 320 Good condition and fully functional
## 369
## biddable startprice condition cellular carrier color storage
## 1 0 105.00 Used 1 AT&T Unknown 32
## 142 1 0.99 Used 0 None Unknown 16
## 309 0 200.00 Used 1 AT&T Black 32
## 312 1 0.99 Used 0 None Unknown 16
## 320 1 60.00 Used 0 None White 16
## 369 1 197.97 Used 0 None Unknown 64
## productline UniqueID
## 1 iPad 1 11862
## 142 iPad mini 12003
## 309 iPad 3 12170
## 312 iPad mini 2 12173
## 320 iPad 1 12181
## 369 iPad mini 3 12230
## description
## 793 Crack on digitizer near top. Top line of digitizer does not respond to touch. Other than that, all
## 794
## 795
## 796
## 797
## 798 Slightly Used. Includes everything you need plus a nice leather case!\nThere is a slice mark on the
## biddable startprice condition cellular carrier color
## 793 0 104.00 For parts or not working 1 Unknown Black
## 794 0 95.00 Used 1 AT&T Unknown
## 795 1 199.99 Manufacturer refurbished 0 None White
## 796 0 149.99 Used 0 None Unknown
## 797 0 7.99 New Unknown Unknown Unknown
## 798 0 139.00 Used 1 Unknown Black
## storage productline UniqueID
## 793 16 iPad 2 12654
## 794 64 iPad 1 12655
## 795 16 iPad 4 12656
## 796 16 iPad 2 12657
## 797 Unknown iPad 3 12658
## 798 32 Unknown 12659
## 'data.frame': 798 obs. of 10 variables:
## $ description: chr "like new" "Item is in great shape. I upgraded to the iPad Air 2 and don't need the mini any longer, even though " "This iPad is working and is tested 100%. It runs great. It is in good condition. Cracked digitizer." "" ...
## $ biddable : int 0 0 0 1 0 0 0 0 0 1 ...
## $ startprice : num 105 195 220 100 211 ...
## $ condition : chr "Used" "Used" "Used" "Used" ...
## $ cellular : chr "1" "0" "0" "0" ...
## $ carrier : chr "AT&T" "None" "None" "None" ...
## $ color : chr "Unknown" "Unknown" "Unknown" "Unknown" ...
## $ storage : chr "32" "16" "64" "16" ...
## $ productline: chr "iPad 1" "iPad mini 2" "iPad 3" "iPad mini" ...
## $ UniqueID : int 11862 11863 11864 11865 11866 11867 11868 11869 11870 11871 ...
## - attr(*, "comment")= chr "glb_newobs_df"
## NULLif ((num_nas <- sum(is.na(glb_trnobs_df[, glb_rsp_var_raw]))) > 0)
stop("glb_trnobs_df$", glb_rsp_var_raw, " contains NAs for ", num_nas, " obs")
if (nrow(glb_trnobs_df) == nrow(glb_allobs_df))
warning("glb_trnobs_df same as glb_allobs_df")
if (nrow(glb_newobs_df) == nrow(glb_allobs_df))
warning("glb_newobs_df same as glb_allobs_df")
if (length(glb_drop_vars) > 0) {
warning("dropping vars: ", paste0(glb_drop_vars, collapse=", "))
glb_allobs_df <- glb_allobs_df[, setdiff(names(glb_allobs_df), glb_drop_vars)]
glb_trnobs_df <- glb_trnobs_df[, setdiff(names(glb_trnobs_df), glb_drop_vars)]
glb_newobs_df <- glb_newobs_df[, setdiff(names(glb_newobs_df), glb_drop_vars)]
}
#stop(here"); sav_allobs_df <- glb_allobs_df # glb_allobs_df <- sav_allobs_df
# Combine trnent & newobs into glb_allobs_df for easier manipulation
glb_trnobs_df$.src <- "Train"; glb_newobs_df$.src <- "Test";
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features, ".src")
glb_allobs_df <- myrbind_df(glb_trnobs_df, glb_newobs_df)
comment(glb_allobs_df) <- "glb_allobs_df"
# Check for duplicates in glb_id_var
if (length(glb_id_var) == 0) {
warning("using .rownames as identifiers for observations")
glb_allobs_df$.rownames <- rownames(glb_allobs_df)
glb_trnobs_df$.rownames <- rownames(subset(glb_allobs_df, .src == "Train"))
glb_newobs_df$.rownames <- rownames(subset(glb_allobs_df, .src == "Test"))
glb_id_var <- ".rownames"
}
if (sum(duplicated(glb_allobs_df[, glb_id_var, FALSE])) > 0)
stop(glb_id_var, " duplicated in glb_allobs_df")
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features, glb_id_var)
glb_allobs_df <- orderBy(reformulate(glb_id_var), glb_allobs_df)
glb_trnobs_df <- glb_newobs_df <- NULL
# For Tableau
write.csv(glb_allobs_df, "data/eBayiPadAll.csv", row.names=FALSE)
#stop(here"); glb_to_sav()
# Make any data corrections here
glb_allobs_df[glb_allobs_df[, glb_id_var] == 10986, "cellular"] <- "1"
glb_allobs_df[glb_allobs_df[, glb_id_var] == 10986, "carrier"] <- "T-Mobile"
# if (!is.null(glb_drop_obs_condition))
# glb_allobs_df <- do.call("subset",
# list(glb_allobs_df, parse(text=paste0("!(", glb_drop_obs_condition, ")"))))
# - Merge glb_obs_stack_condition & glb_obs_drop_condition
# - Derive glb_obs_stack|drop_chk_vars from condition automatically
# - Implement glb_obs_stack_condition & glb_obs_stack_chk_vars options
dsp_partition_stats <- function(obs_df, vars=NULL) {
lcl_vars <- NULL
for (var in c(vars, glb_rsp_var_raw)) {
if ((length(unique(obs_df[, var])) > 5) && is.numeric(obs_df[, var])) {
cut_var <- paste0(var, ".cut.fctr")
obs_df[, cut_var] <- cut(obs_df[, var], 3)
lcl_vars <- union(lcl_vars, cut_var)
} else lcl_vars <- union(lcl_vars, var)
}
print("Partition stats:")
print(mycreate_sqlxtab_df(obs_df, union(lcl_vars, ".src")))
for (var in lcl_vars) {
print(freq_df <- mycreate_sqlxtab_df(obs_df, union(var, ".src")))
print(myplot_hbar(freq_df, ".src", ".n", colorcol_name=var))
}
print(mycreate_sqlxtab_df(obs_df, ".src"))
# if (length(unique(glb_allobs_df[, glb_rsp_var_raw])) > 5) {
# cut_var <- paste0(glb_rsp_var_raw, ".cut.fctr")
# glb_allobs_df[, cut_var] <- cut(glb_allobs_df[, glb_rsp_var_raw], 3)
# glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features, cut_var)
# glb_obs_stack_chk_vars <- union(cut_var, glb_obs_stack_chk_vars)
# } else glb_obs_stack_chk_vars <- union(glb_rsp_var_raw, glb_obs_stack_chk_vars)
# #glb_obs_stack_chk_vars <- union(glb_obs_stack_chk_vars, ".src")
# print(mycreate_sqlxtab_df(glb_allobs_df, union(var, ".src")))
# print(mycreate_sqlxtab_df(glb_allobs_df, union(glb_obs_stack_chk_vars, ".src")))
# for (var in glb_obs_stack_chk_vars) {
# print(mycreate_sqlxtab_df(glb_allobs_df, union(var, ".src")))
# }
# print(mycreate_sqlxtab_df(glb_allobs_df, ".src"))
}
myget_symbols <- function(txt) {
if (is.null(txt)) return(NULL)
#print(getParseData(parse(text=txt, keep.source=TRUE)))
return(unique(subset(getParseData(parse(text=txt, keep.source=TRUE)),
token == "SYMBOL")$text))
}
# tokens <- unlist(strsplit(gsub("[[:punct:]|[:space:]]", " ", glb_obs_drop_condition), " "))
# tokens <- tokens[tokens != ""]
# glb_obs_drop_chk_vars <- c("biddable") # or NULL
dsp_partition_stats(obs_df=glb_allobs_df, vars=myget_symbols(glb_obs_drop_condition))## [1] "Partition stats:"## Loading required package: sqldf
## Loading required package: gsubfn
## Loading required package: proto
## Loading required package: RSQLite
## Loading required package: DBI
## Loading required package: tcltk## UniqueID.cut.fctr sold .src .n
## 1 (1.18e+04,1.27e+04] NA Test 798
## 2 (1.09e+04,1.18e+04] 0 Train 589
## 3 (1e+04,1.09e+04] 1 Train 531
## 4 (1e+04,1.09e+04] 0 Train 356
## 5 (1.09e+04,1.18e+04] 1 Train 297
## 6 (1.18e+04,1.27e+04] 0 Train 56
## 7 (1.18e+04,1.27e+04] 1 Train 32
## UniqueID.cut.fctr .src .n
## 1 (1e+04,1.09e+04] Train 887
## 2 (1.09e+04,1.18e+04] Train 886
## 3 (1.18e+04,1.27e+04] Test 798
## 4 (1.18e+04,1.27e+04] Train 88
## sold .src .n
## 1 0 Train 1001
## 2 1 Train 860
## 3 NA Test 798
## .src .n
## 1 Train 1861
## 2 Test 798if (!is.null(glb_obs_drop_condition)) {
print(sprintf("Running glb_obs_drop_condition filter: %s", glb_obs_drop_condition))
glb_allobs_df <- do.call("subset",
list(glb_allobs_df, parse(text=paste0("!(", glb_obs_drop_condition, ")"))))
dsp_partition_stats(obs_df=glb_allobs_df, vars=myget_symbols(glb_obs_drop_condition))
}## [1] "Running glb_obs_drop_condition filter: (UniqueID %in% c(NULL\n , 11234 #sold=0; 2 other dups(10306, 11503) are sold=1\n , 11844 #sold=0; 3 other dups(11721, 11738, 11812) are sold=1\n )) \n # | (productline %in% c('iPad 5', 'iPad mini Retina'))\n # | (biddable != 0) # bid0_sp\n # | (biddable == 0) # bid1_sp\n "
## [1] "Partition stats:"
## UniqueID.cut.fctr sold .src .n
## 1 (1.18e+04,1.27e+04] NA Test 798
## 2 (1.09e+04,1.18e+04] 0 Train 588
## 3 (1e+04,1.09e+04] 1 Train 531
## 4 (1e+04,1.09e+04] 0 Train 356
## 5 (1.09e+04,1.18e+04] 1 Train 297
## 6 (1.18e+04,1.27e+04] 0 Train 55
## 7 (1.18e+04,1.27e+04] 1 Train 32
## UniqueID.cut.fctr .src .n
## 1 (1e+04,1.09e+04] Train 887
## 2 (1.09e+04,1.18e+04] Train 885
## 3 (1.18e+04,1.27e+04] Test 798
## 4 (1.18e+04,1.27e+04] Train 87
## sold .src .n
## 1 0 Train 999
## 2 1 Train 860
## 3 NA Test 798
## .src .n
## 1 Train 1859
## 2 Test 798# Check for duplicates by all features
require(gdata)## Loading required package: gdata
## gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
##
## gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
##
## Attaching package: 'gdata'
##
## The following object is masked from 'package:stats':
##
## nobs
##
## The following object is masked from 'package:utils':
##
## object.size#print(names(glb_allobs_df))
dup_allobs_df <- glb_allobs_df[duplicated2(subset(glb_allobs_df,
select=-c(UniqueID, sold, .src))), ]
dup_allobs_df <- orderBy(~productline+description+startprice+biddable, dup_allobs_df)
print(sprintf("Found %d duplicates by all features:", nrow(dup_allobs_df)))## [1] "Found 304 duplicates by all features:"myprint_df(dup_allobs_df)## description biddable startprice condition cellular
## 1711 1 0.99 For parts or not working Unknown
## 2608 1 0.99 For parts or not working Unknown
## 293 1 5.00 Used Unknown
## 478 1 5.00 Used Unknown
## 385 0 15.00 Used 0
## 390 0 15.00 Used 0
## carrier color storage productline sold UniqueID .src
## 1711 Unknown Unknown 16 Unknown 1 11711 Train
## 2608 Unknown Unknown 16 Unknown NA 12608 Test
## 293 Unknown White 16 Unknown 1 10293 Train
## 478 Unknown White 16 Unknown 1 10478 Train
## 385 None Black 16 Unknown 0 10385 Train
## 390 None Black 16 Unknown 0 10390 Train
## description biddable startprice condition cellular
## 1956 1 0.99 Used 0
## 828 1 249.97 Manufacturer refurbished 1
## 3 0 199.99 Used 0
## 1649 0 209.00 For parts or not working Unknown
## 2111 1 200.00 Used 0
## 172 0 269.00 Used 0
## carrier color storage productline sold UniqueID .src
## 1956 None Unknown 16 iPad 2 NA 11956 Test
## 828 Unknown Black 64 iPad 2 0 10828 Train
## 3 None White 16 iPad 4 1 10003 Train
## 1649 Unknown Unknown 16 iPad Air 0 11649 Train
## 2111 None Space Gray 64 iPad mini 2 NA 12111 Test
## 172 None Unknown 32 iPad mini 2 0 10172 Train
## description biddable startprice condition cellular carrier color
## 8 0 329.99 New 0 None White
## 660 0 329.99 New 0 None White
## 319 0 345.00 New 0 None Gold
## 1886 0 345.00 New 0 None Gold
## 1363 0 498.88 New 1 Verizon Gold
## 1394 0 498.88 New 1 Verizon Gold
## storage productline sold UniqueID .src
## 8 16 iPad mini 3 0 10008 Train
## 660 16 iPad mini 3 0 10660 Train
## 319 16 iPad mini 3 1 10319 Train
## 1886 16 iPad mini 3 NA 11886 Test
## 1363 16 iPad mini 3 0 11363 Train
## 1394 16 iPad mini 3 0 11394 Train# print(dup_allobs_df[, c(glb_id_var, glb_rsp_var_raw,
# "description", "startprice", "biddable")])
# write.csv(dup_allobs_df[, c("UniqueID"), FALSE], "ebayipads_dups.csv", row.names=FALSE)
dupobs_df <- tidyr::unite(dup_allobs_df, "allfeats", -c(sold, UniqueID, .src), sep="#")
# dupobs_df <- dplyr::group_by(dupobs_df, allfeats)
# dupobs_df <- dupobs_df[, "UniqueID", FALSE]
# dupobs_df <- ungroup(dupobs_df)
#
# dupobs_df$.rownames <- row.names(dupobs_df)
grpobs_df <- data.frame(allfeats=unique(dupobs_df[, "allfeats"]))
grpobs_df$.grpid <- row.names(grpobs_df)
dupobs_df <- merge(dupobs_df, grpobs_df)
# dupobs_tbl <- table(dupobs_df$.grpid)
# print(max(dupobs_tbl))
# print(dupobs_tbl[which.max(dupobs_tbl)])
# print(dupobs_df[dupobs_df$.grpid == names(dupobs_tbl[which.max(dupobs_tbl)]), ])
# print(dupobs_df[dupobs_df$.grpid == 106, ])
# for (grpid in c(9, 17, 31, 36, 53))
# print(dupobs_df[dupobs_df$.grpid == grpid, ])
dupgrps_df <- as.data.frame(table(dupobs_df$.grpid, dupobs_df$sold, useNA="ifany"))
names(dupgrps_df)[c(1,2)] <- c(".grpid", "sold")
dupgrps_df$.grpid <- as.numeric(as.character(dupgrps_df$.grpid))
dupgrps_df <- tidyr::spread(dupgrps_df, sold, Freq)
names(dupgrps_df)[-1] <- paste("sold", names(dupgrps_df)[-1], sep=".")
dupgrps_df$.freq <- sapply(1:nrow(dupgrps_df), function(row) sum(dupgrps_df[row, -1]))
myprint_df(orderBy(~-.freq, dupgrps_df))## .grpid sold.0 sold.1 sold.NA .freq
## 40 40 0 6 3 9
## 106 106 0 4 1 5
## 9 9 0 1 3 4
## 17 17 0 3 1 4
## 36 36 0 3 1 4
## 53 53 0 2 2 4
## .grpid sold.0 sold.1 sold.NA .freq
## 10 10 0 2 0 2
## 42 42 0 1 1 2
## 57 57 1 0 1 2
## 66 66 1 0 1 2
## 91 91 0 1 1 2
## 101 101 0 1 1 2
## .grpid sold.0 sold.1 sold.NA .freq
## 130 130 1 0 1 2
## 131 131 1 1 0 2
## 132 132 0 1 1 2
## 133 133 2 0 0 2
## 134 134 0 1 1 2
## 135 135 2 0 0 2print("sold Conflicts:")## [1] "sold Conflicts:"print(subset(dupgrps_df, (sold.0 > 0) & (sold.1 > 0)))## .grpid sold.0 sold.1 sold.NA .freq
## 4 4 1 1 0 2
## 22 22 1 1 0 2
## 23 23 1 1 0 2
## 74 74 1 1 0 2
## 83 83 1 1 0 2
## 84 84 1 1 0 2
## 95 95 1 1 0 2
## 102 102 1 1 0 2
## 109 109 1 1 0 2
## 111 111 1 1 0 2
## 122 122 1 1 0 2
## 131 131 1 1 0 2#dupobs_df[dupobs_df$.grpid == 4, ]
glb_allobs_df <- merge(glb_allobs_df, dupobs_df[, c(glb_id_var, ".grpid")],
by=glb_id_var, all.x=TRUE)
if (nrow(subset(dupgrps_df, (sold.0 > 0) & (sold.1 > 0) & (sold.0 != sold.1))) > 0)
stop("Duplicate conflicts are resolvable")
#subset(glb_allobs_df, .grpid %in% c(25))
#mydsp_obs(list(productline.contains="iPad 1", storage.contains="16", color.contains="Black", carrier.contains="None", cellular.contains="0", condition.contains="Used", startprice=80), cols=c("productline", "storage", "color", "carrier", "cellular", "condition", "startprice", "sold"))
print("Test & Train Groups:")## [1] "Test & Train Groups:"print(subset(dupgrps_df, (sold.NA > 0)))## .grpid sold.0 sold.1 sold.NA .freq
## 1 1 0 1 1 2
## 5 5 1 0 1 2
## 7 7 0 0 2 2
## 8 8 1 0 1 2
## 9 9 0 1 3 4
## 12 12 0 0 2 2
## 14 14 0 1 1 2
## 15 15 0 0 2 2
## 17 17 0 3 1 4
## 18 18 0 2 1 3
## 19 19 0 2 1 3
## 24 24 0 2 1 3
## 26 26 1 0 1 2
## 28 28 1 0 1 2
## 30 30 0 1 1 2
## 32 32 0 0 2 2
## 33 33 0 1 1 2
## 35 35 0 2 1 3
## 36 36 0 3 1 4
## 37 37 0 0 2 2
## 38 38 0 1 1 2
## 40 40 0 6 3 9
## 41 41 0 0 2 2
## 42 42 0 1 1 2
## 43 43 0 1 1 2
## 44 44 0 2 1 3
## 47 47 0 1 1 2
## 48 48 0 0 2 2
## 49 49 0 1 2 3
## 51 51 0 1 1 2
## 53 53 0 2 2 4
## 54 54 0 1 1 2
## 55 55 1 0 2 3
## 56 56 1 0 1 2
## 57 57 1 0 1 2
## 58 58 0 0 2 2
## 59 59 1 0 1 2
## 60 60 1 0 1 2
## 63 63 0 1 1 2
## 66 66 1 0 1 2
## 67 67 1 0 1 2
## 68 68 0 0 2 2
## 69 69 1 0 1 2
## 73 73 0 1 1 2
## 76 76 0 2 1 3
## 86 86 0 0 2 2
## 87 87 1 0 1 2
## 89 89 1 0 1 2
## 90 90 0 0 2 2
## 91 91 0 1 1 2
## 93 93 0 1 1 2
## 94 94 1 0 1 2
## 99 99 0 1 1 2
## 101 101 0 1 1 2
## 103 103 0 1 1 2
## 104 104 1 0 1 2
## 106 106 0 4 1 5
## 107 107 0 1 1 2
## 108 108 0 1 1 2
## 112 112 1 0 1 2
## 114 114 0 1 1 2
## 115 115 0 1 1 2
## 116 116 1 0 1 2
## 117 117 0 2 1 3
## 118 118 0 1 1 2
## 121 121 1 0 1 2
## 124 124 1 0 1 2
## 128 128 0 1 1 2
## 130 130 1 0 1 2
## 132 132 0 1 1 2
## 134 134 0 1 1 2glb_exclude_vars_as_features <- c(".grpid", glb_exclude_vars_as_features)
#stop("Implement code for glb_inp_merge_lst")
# spd_allobs_df <- read.csv(paste0(glb_out_pfx, "sp_predict.csv"))
# if (nrow(spd_allobs_df) != nrow(glb_allobs_df))
# stop("mismatches between spd_allobs_df & glb_allobs_df")
# mrg_allobs_df <- merge(glb_allobs_df, spd_allobs_df)
# if (nrow(mrg_allobs_df) != nrow(glb_allobs_df))
# stop("mismatches between mrg_allobs_df & glb_allobs_df")
# mrg_allobs_df$startprice.diff <- mrg_allobs_df$startprice -
# (exp(mrg_allobs_df$startprice.log.predict.) - 1)
# mrg_allobs_df$startprice.log.diff <- log(1 + mrg_allobs_df$startprice) -
# mrg_allobs_df$startprice.log.predict.
# print(myplot_scatter(mrg_allobs_df, "startprice", "startprice.diff",
# colorcol_name = "biddable"))
# print(myplot_scatter(mrg_allobs_df, "startprice", "startprice.log.diff",
# colorcol_name = "biddable"))
# print(myplot_histogram(mrg_allobs_df, "startprice.diff",
# fill_col_name = "biddable"))
# print(myplot_histogram(mrg_allobs_df, "startprice.log.diff",
# fill_col_name = "biddable"))
# glb_allobs_df <- mrg_allobs_df
# glb_exclude_vars_as_features <- c(glb_exclude_vars_as_features,
# "startprice.diff", "startprice.log", "startprice.log.predict.")
#stop(here"); glb_to_sav(); all.equal(sav_allobs_df, glb_allobs_df); glb_allobs_df <- sav_allobs_df
dsp_partition_stats(obs_df=glb_allobs_df,
vars=myget_symbols(glb_obs_repartition_train_condition))## [1] "Partition stats:"
## sold .src .n
## 1 0 Train 999
## 2 1 Train 860
## 3 NA Test 798
## sold .src .n
## 1 0 Train 999
## 2 1 Train 860
## 3 NA Test 798
## .src .n
## 1 Train 1859
## 2 Test 798if (!is.null(glb_obs_repartition_train_condition)) {
print(sprintf("Running glb_obs_repartition_train_condition filter: %s",
glb_obs_repartition_train_condition))
# glb_allobs_df <- mutate(glb_allobs_df, .src=ifelse(!is.na(sold) & (sold == 1),
# "Train", "Test"))
# glb_allobs_df <- mutate_(glb_allobs_df,
# .src=interp(ifelse(eval(parse(text="!is.na(sold) & (sold == 1)")),
# "Train", "Test")))
# glb_allobs_df <- within(glb_allobs_df, {
# .src <- ifelse(eval(parse(text="!is.na(sold) & (sold == 1)")),
# "Train", "Test")
# })
# glb_allobs_df <- within(glb_allobs_df, {
# if(eval(parse(text="!is.na(sold) & (sold == 1)"))) .src <- "Train" else
# .src <- "Test"
# })
# with(glb_allobs_df, {
# src <- ifelse(eval(parse(text="!is.na(sold) & (sold == 1)")),
# "Train", "Test")
# })
# glb_allobs_df$.src <- sapply(1:nrow(glb_allobs_df), function (row_ix) ifelse)
# glb_allobs_df[parse(text=paste0("!(", glb_obs_drop_condition, ")")), ".src"] <- do.call("subset",
# list(glb_allobs_df, ))
glb_trnobs_df <- do.call("subset", list(glb_allobs_df,
parse(text=paste0(" (", glb_obs_repartition_train_condition, ")"))))
glb_trnobs_df$.src <- "Train"
glb_newobs_df <- do.call("subset", list(glb_allobs_df,
parse(text=paste0("!(", glb_obs_repartition_train_condition, ")"))))
glb_newobs_df$.src <- "Test"
glb_allobs_df <- rbind(glb_trnobs_df, glb_newobs_df)
dsp_partition_stats(obs_df=glb_allobs_df,
vars=myget_symbols(glb_obs_repartition_train_condition))
}
glb_chunks_df <- myadd_chunk(glb_chunks_df, "inspect.data", major.inc=TRUE)## label step_major step_minor label_minor bgn end elapsed
## 1 import.data 1 0 0 11.272 22.337 11.065
## 2 inspect.data 2 0 0 22.338 NA NAStep 2.0: inspect data
#print(str(glb_allobs_df))
#View(glb_allobs_df)
dsp_class_dstrb <- function(var) {
xtab_df <- mycreate_xtab_df(glb_allobs_df, c(".src", var))
rownames(xtab_df) <- xtab_df$.src
xtab_df <- subset(xtab_df, select=-.src)
print(xtab_df)
print(xtab_df / rowSums(xtab_df, na.rm=TRUE))
}
# Performed repeatedly in other chunks
glb_chk_data <- function(fctrMaxUniqVals = glbFctrMaxUniqVals) {
# Histogram of predictor in glb_trnobs_df & glb_newobs_df
print(myplot_histogram(glb_allobs_df, glb_rsp_var_raw) + facet_wrap(~ .src))
if (glb_is_classification)
dsp_class_dstrb(var=ifelse(glb_rsp_var %in% names(glb_allobs_df),
glb_rsp_var, glb_rsp_var_raw))
mycheck_problem_data(glb_allobs_df, fctrMaxUniqVals)
}
glb_chk_data()## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.## Warning: Removed 798 rows containing non-finite values (stat_bin).## Loading required package: reshape2
## sold.0 sold.1 sold.NA
## Test NA NA 798
## Train 999 860 NA
## sold.0 sold.1 sold.NA
## Test NA NA 1
## Train 0.5373857 0.4626143 NA
## [1] "numeric data missing in : "
## sold
## 798
## [1] "numeric data w/ 0s in : "
## biddable sold
## 1444 999
## [1] "numeric data w/ Infs in : "
## named integer(0)
## [1] "numeric data w/ NaNs in : "
## named integer(0)
## [1] "string data missing in : "
## description condition cellular carrier color storage
## 1520 0 0 0 0 0
## productline .grpid
## 0 NA# Create new features that help diagnostics
if (!is.null(glb_map_rsp_raw_to_var)) {
glb_allobs_df[, glb_rsp_var] <-
glb_map_rsp_raw_to_var(glb_allobs_df[, glb_rsp_var_raw])
mycheck_map_results(mapd_df=glb_allobs_df,
from_col_name=glb_rsp_var_raw, to_col_name=glb_rsp_var)
if (glb_is_classification) dsp_class_dstrb(glb_rsp_var)
}## sold sold.fctr .n
## 1 0 N 999
## 2 1 Y 860
## 3 NA <NA> 798## Warning: Removed 1 rows containing missing values (position_stack).
## sold.fctr.N sold.fctr.Y sold.fctr.NA
## Test NA NA 798
## Train 999 860 NA
## sold.fctr.N sold.fctr.Y sold.fctr.NA
## Test NA NA 1
## Train 0.5373857 0.4626143 NA# check distribution of all numeric data
dsp_numeric_feats_dstrb <- function(feats_vctr) {
for (feat in feats_vctr) {
print(sprintf("feat: %s", feat))
if (glb_is_regression)
gp <- myplot_scatter(df=glb_allobs_df, ycol_name=glb_rsp_var, xcol_name=feat,
smooth=TRUE)
if (glb_is_classification)
gp <- myplot_box(df=glb_allobs_df, ycol_names=feat, xcol_name=glb_rsp_var)
if (inherits(glb_allobs_df[, feat], "factor"))
gp <- gp + facet_wrap(reformulate(feat))
print(gp)
}
}
# dsp_numeric_vars_dstrb(setdiff(names(glb_allobs_df),
# union(myfind_chr_cols_df(glb_allobs_df),
# c(glb_rsp_var_raw, glb_rsp_var))))
add_new_diag_feats <- function(obs_df, ref_df=glb_allobs_df) {
require(plyr)
set.seed(169)
obs_df <- mutate(obs_df,
# <col_name>.NA=is.na(<col_name>),
# <col_name>.fctr=factor(<col_name>,
# as.factor(union(obs_df$<col_name>, obs_twin_df$<col_name>))),
# <col_name>.fctr=relevel(factor(<col_name>,
# as.factor(union(obs_df$<col_name>, obs_twin_df$<col_name>))),
# "<ref_val>"),
# <col2_name>.fctr=relevel(factor(ifelse(<col1_name> == <val>, "<oth_val>", "<ref_val>")),
# as.factor(c("R", "<ref_val>")),
# ref="<ref_val>"),
# This doesn't work - use sapply instead
# <col_name>.fctr_num=grep(<col_name>, levels(<col_name>.fctr)),
#
# Date.my=as.Date(strptime(Date, "%m/%d/%y %H:%M")),
# Year=year(Date.my),
# Month=months(Date.my),
# Weekday=weekdays(Date.my)
# <col_name>=<table>[as.character(<col2_name>)],
# <col_name>=as.numeric(<col2_name>),
# <col_name> = trunc(<col2_name> / 100),
.rnorm = rnorm(n=nrow(obs_df))
)
# If levels of a factor are different across obs_df & glb_newobs_df; predict.glm fails
# Transformations not handled by mutate
# obs_df$<col_name>.fctr.num <- sapply(1:nrow(obs_df),
# function(row_ix) grep(obs_df[row_ix, "<col_name>"],
# levels(obs_df[row_ix, "<col_name>.fctr"])))
#print(summary(obs_df))
#print(sapply(names(obs_df), function(col) sum(is.na(obs_df[, col]))))
return(obs_df)
}
glb_allobs_df <- add_new_diag_feats(glb_allobs_df)## Loading required package: plyrrequire(dplyr)## Loading required package: dplyr
##
## Attaching package: 'dplyr'
##
## The following objects are masked from 'package:plyr':
##
## arrange, count, desc, failwith, id, mutate, rename, summarise,
## summarize
##
## The following objects are masked from 'package:gdata':
##
## combine, first, last
##
## The following objects are masked from 'package:stats':
##
## filter, lag
##
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union#stop(here"); sav_allobs_df <- glb_allobs_df # glb_allobs_df <- sav_allobs_df
# Merge some <descriptor>
# glb_allobs_df$<descriptor>.my <- glb_allobs_df$<descriptor>
# glb_allobs_df[grepl("\\bAIRPORT\\b", glb_allobs_df$<descriptor>.my),
# "<descriptor>.my"] <- "AIRPORT"
# Check distributions of newly transformed / extracted vars
# Enhancement: remove vars that were displayed ealier
dsp_numeric_feats_dstrb(feats_vctr=setdiff(names(glb_allobs_df),
c(myfind_chr_cols_df(glb_allobs_df), glb_rsp_var_raw, glb_rsp_var,
glb_exclude_vars_as_features)))## [1] "feat: biddable"
## [1] "feat: startprice"
## [1] "feat: .rnorm"
# Convert factors to dummy variables
# Build splines require(splines); bsBasis <- bs(training$age, df=3)
#pairs(subset(glb_trnobs_df, select=-c(col_symbol)))
# Check for glb_newobs_df & glb_trnobs_df features range mismatches
# Other diagnostics:
# print(subset(glb_trnobs_df, <col1_name> == max(glb_trnobs_df$<col1_name>, na.rm=TRUE) &
# <col2_name> <= mean(glb_trnobs_df$<col1_name>, na.rm=TRUE)))
# print(glb_trnobs_df[which.max(glb_trnobs_df$<col_name>),])
# print(<col_name>_freq_glb_trnobs_df <- mycreate_tbl_df(glb_trnobs_df, "<col_name>"))
# print(which.min(table(glb_trnobs_df$<col_name>)))
# print(which.max(table(glb_trnobs_df$<col_name>)))
# print(which.max(table(glb_trnobs_df$<col1_name>, glb_trnobs_df$<col2_name>)[, 2]))
# print(table(glb_trnobs_df$<col1_name>, glb_trnobs_df$<col2_name>))
# print(table(is.na(glb_trnobs_df$<col1_name>), glb_trnobs_df$<col2_name>))
# print(table(sign(glb_trnobs_df$<col1_name>), glb_trnobs_df$<col2_name>))
# print(mycreate_xtab_df(glb_trnobs_df, <col1_name>))
# print(mycreate_xtab_df(glb_trnobs_df, c(<col1_name>, <col2_name>)))
# print(<col1_name>_<col2_name>_xtab_glb_trnobs_df <-
# mycreate_xtab_df(glb_trnobs_df, c("<col1_name>", "<col2_name>")))
# <col1_name>_<col2_name>_xtab_glb_trnobs_df[is.na(<col1_name>_<col2_name>_xtab_glb_trnobs_df)] <- 0
# print(<col1_name>_<col2_name>_xtab_glb_trnobs_df <-
# mutate(<col1_name>_<col2_name>_xtab_glb_trnobs_df,
# <col3_name>=(<col1_name> * 1.0) / (<col1_name> + <col2_name>)))
# print(mycreate_sqlxtab_df(glb_allobs_df, c("<col1_name>", "<col2_name>")))
# print(<col2_name>_min_entity_arr <-
# sort(tapply(glb_trnobs_df$<col1_name>, glb_trnobs_df$<col2_name>, min, na.rm=TRUE)))
# print(<col1_name>_na_by_<col2_name>_arr <-
# sort(tapply(glb_trnobs_df$<col1_name>.NA, glb_trnobs_df$<col2_name>, mean, na.rm=TRUE)))
# Other plots:
# print(myplot_box(df=glb_trnobs_df, ycol_names="<col1_name>"))
# print(myplot_box(df=glb_trnobs_df, ycol_names="<col1_name>", xcol_name="<col2_name>"))
# print(myplot_line(subset(glb_trnobs_df, Symbol %in% c("CocaCola", "ProcterGamble")),
# "Date.POSIX", "StockPrice", facet_row_colnames="Symbol") +
# geom_vline(xintercept=as.numeric(as.POSIXlt("2003-03-01"))) +
# geom_vline(xintercept=as.numeric(as.POSIXlt("1983-01-01")))
# )
# print(myplot_line(subset(glb_trnobs_df, Date.POSIX > as.POSIXct("2004-01-01")),
# "Date.POSIX", "StockPrice") +
# geom_line(aes(color=Symbol)) +
# coord_cartesian(xlim=c(as.POSIXct("1990-01-01"),
# as.POSIXct("2000-01-01"))) +
# coord_cartesian(ylim=c(0, 250)) +
# geom_vline(xintercept=as.numeric(as.POSIXlt("1997-09-01"))) +
# geom_vline(xintercept=as.numeric(as.POSIXlt("1997-11-01")))
# )
# print(myplot_scatter(glb_allobs_df, "<col1_name>", "<col2_name>", smooth=TRUE))
# print(myplot_scatter(glb_allobs_df, "<col1_name>", "<col2_name>", colorcol_name="<Pred.fctr>") +
# geom_point(data=subset(glb_allobs_df, <condition>),
# mapping=aes(x=<x_var>, y=<y_var>), color="red", shape=4, size=5) +
# geom_vline(xintercept=84))
glb_chunks_df <- myadd_chunk(glb_chunks_df, "scrub.data", major.inc=FALSE)## label step_major step_minor label_minor bgn end elapsed
## 2 inspect.data 2 0 0 22.338 26.486 4.149
## 3 scrub.data 2 1 1 26.487 NA NAStep 2.1: scrub data
mycheck_problem_data(glb_allobs_df, fctrMaxUniqVals = glbFctrMaxUniqVals)## [1] "numeric data missing in : "
## sold sold.fctr
## 798 798
## [1] "numeric data w/ 0s in : "
## biddable sold
## 1444 999
## [1] "numeric data w/ Infs in : "
## named integer(0)
## [1] "numeric data w/ NaNs in : "
## named integer(0)
## [1] "string data missing in : "
## description condition cellular carrier color storage
## 1520 0 0 0 0 0
## productline .grpid
## 0 NAfindOffendingCharacter <- function(x, maxStringLength=256){
print(x)
for (c in 1:maxStringLength){
offendingChar <- substr(x,c,c)
#print(offendingChar) #uncomment if you want the indiv characters printed
#the next character is the offending multibyte Character
}
}
# string_vector <- c("test", "Se\x96ora", "works fine")
# lapply(string_vector, findOffendingCharacter)
# lapply(glb_allobs_df$description[29], findOffendingCharacter)
dsp_hdlxtab <- function(str)
print(mycreate_sqlxtab_df(glb_allobs_df[sel_obs(Headline.contains=str), ],
c("Headline.pfx", "Headline", glb_rsp_var)))
#dsp_hdlxtab("(1914)|(1939)")
dsp_catxtab <- function(str)
print(mycreate_sqlxtab_df(glb_allobs_df[sel_obs(Headline.contains=str), ],
c("Headline.pfx", "NewsDesk", "SectionName", "SubsectionName", glb_rsp_var)))
# dsp_catxtab("1914)|(1939)")
# dsp_catxtab("19(14|39|64):")
# dsp_catxtab("19..:")
# Merge some categories
# glb_allobs_df$myCategory <-
# plyr::revalue(glb_allobs_df$myCategory, c(
# "#Business Day#Dealbook" = "Business#Business Day#Dealbook",
# "#Business Day#Small Business" = "Business#Business Day#Small Business",
# "dummy" = "dummy"
# ))
# ctgry_xtab_df <- orderBy(reformulate(c("-", ".n")),
# mycreate_sqlxtab_df(glb_allobs_df,
# c("myCategory", "NewsDesk", "SectionName", "SubsectionName", glb_rsp_var)))
# myprint_df(ctgry_xtab_df)
# write.table(ctgry_xtab_df, paste0(glb_out_pfx, "ctgry_xtab.csv"),
# row.names=FALSE)
# ctgry_cast_df <- orderBy(~ -Y -NA, dcast(ctgry_xtab_df,
# myCategory + NewsDesk + SectionName + SubsectionName ~
# Popular.fctr, sum, value.var=".n"))
# myprint_df(ctgry_cast_df)
# write.table(ctgry_cast_df, paste0(glb_out_pfx, "ctgry_cast.csv"),
# row.names=FALSE)
# print(ctgry_sum_tbl <- table(glb_allobs_df$myCategory, glb_allobs_df[, glb_rsp_var],
# useNA="ifany"))
dsp_chisq.test <- function(...) {
sel_df <- glb_allobs_df[sel_obs(...) &
!is.na(glb_allobs_df$Popular), ]
sel_df$.marker <- 1
ref_df <- glb_allobs_df[!is.na(glb_allobs_df$Popular), ]
mrg_df <- merge(ref_df[, c(glb_id_var, "Popular")],
sel_df[, c(glb_id_var, ".marker")], all.x=TRUE)
mrg_df[is.na(mrg_df)] <- 0
print(mrg_tbl <- table(mrg_df$.marker, mrg_df$Popular))
print("Rows:Selected; Cols:Popular")
#print(mrg_tbl)
print(chisq.test(mrg_tbl))
}
# dsp_chisq.test(Headline.contains="[Ee]bola")
# dsp_chisq.test(Snippet.contains="[Ee]bola")
# dsp_chisq.test(Abstract.contains="[Ee]bola")
# print(mycreate_sqlxtab_df(glb_allobs_df[sel_obs(Headline.contains="[Ee]bola"), ],
# c(glb_rsp_var, "NewsDesk", "SectionName", "SubsectionName")))
# print(table(glb_allobs_df$NewsDesk, glb_allobs_df$SectionName))
# print(table(glb_allobs_df$SectionName, glb_allobs_df$SubsectionName))
# print(table(glb_allobs_df$NewsDesk, glb_allobs_df$SectionName, glb_allobs_df$SubsectionName))
# glb_allobs_df$myCategory.fctr <- as.factor(glb_allobs_df$myCategory)
print(table(glb_allobs_df$cellular, glb_allobs_df$carrier, useNA="ifany"))##
## AT&T None Other Sprint T-Mobile Unknown Verizon
## 0 0 1593 0 0 0 0 0
## 1 288 0 4 36 28 172 196
## Unknown 4 4 2 0 0 330 0# glb_allobs_df[(glb_allobs_df$cellular %in% c("Unknown")) &
# (glb_allobs_df$carrier %in% c("AT&T", "Other")),
# c(glb_id_var, glb_rsp_var_raw, "description", "carrier", "cellular")]
glb_allobs_df[(glb_allobs_df$cellular %in% c("Unknown")) &
(glb_allobs_df$carrier %in% c("AT&T", "Other")),
"cellular"] <- "1"
# glb_allobs_df[(glb_allobs_df$cellular %in% c("Unknown")) &
# (glb_allobs_df$carrier %in% c("None")),
# c(glb_id_var, glb_rsp_var_raw, "description", "carrier", "cellular")]
glb_allobs_df[(glb_allobs_df$cellular %in% c("Unknown")) &
(glb_allobs_df$carrier %in% c("None")),
"cellular"] <- "0"
print(table(glb_allobs_df$cellular, glb_allobs_df$carrier, useNA="ifany"))##
## AT&T None Other Sprint T-Mobile Unknown Verizon
## 0 0 1597 0 0 0 0 0
## 1 292 0 6 36 28 172 196
## Unknown 0 0 0 0 0 330 0Step 2.1: scrub data
glb_chunks_df <- myadd_chunk(glb_chunks_df, "transform.data", major.inc=FALSE)## label step_major step_minor label_minor bgn end elapsed
## 3 scrub.data 2 1 1 26.487 27.672 1.185
## 4 transform.data 2 2 2 27.673 NA NA### Mapping dictionary
#sav_allobs_df <- glb_allobs_df; glb_allobs_df <- sav_allobs_df
if (!is.null(glb_map_vars)) {
for (feat in glb_map_vars) {
map_df <- myimport_data(url=glb_map_urls[[feat]],
comment="map_df",
print_diagn=TRUE)
glb_allobs_df <- mymap_codes(glb_allobs_df, feat, names(map_df)[2],
map_df, map_join_col_name=names(map_df)[1],
map_tgt_col_name=names(map_df)[2])
}
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features, glb_map_vars)
}
### Forced Assignments
#stop(here"); sav_allobs_df <- glb_allobs_df; glb_allobs_df <- sav_allobs_df
for (feat in glb_assign_vars) {
new_feat <- paste0(feat, ".my")
print(sprintf("Forced Assignments for: %s -> %s...", feat, new_feat))
glb_allobs_df[, new_feat] <- glb_allobs_df[, feat]
pairs <- glb_assign_pairs_lst[[feat]]
for (pair_ix in 1:length(pairs$from)) {
if (is.na(pairs$from[pair_ix]))
nobs <- nrow(filter(glb_allobs_df,
is.na(eval(parse(text=feat),
envir=glb_allobs_df)))) else
nobs <- sum(glb_allobs_df[, feat] == pairs$from[pair_ix])
#nobs <- nrow(filter(glb_allobs_df, is.na(Married.fctr))) ; print(nobs)
if ((is.na(pairs$from[pair_ix])) && (is.na(pairs$to[pair_ix])))
stop("what are you trying to do ???")
if (is.na(pairs$from[pair_ix]))
glb_allobs_df[is.na(glb_allobs_df[, feat]), new_feat] <-
pairs$to[pair_ix] else
glb_allobs_df[glb_allobs_df[, feat] == pairs$from[pair_ix], new_feat] <-
pairs$to[pair_ix]
print(sprintf(" %s -> %s for %s obs",
pairs$from[pair_ix], pairs$to[pair_ix], format(nobs, big.mark=",")))
}
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features, glb_assign_vars)
}
### Derivations using mapping functions
#stop(here"); sav_allobs_df <- glb_allobs_df; glb_allobs_df <- sav_allobs_df
for (new_feat in glb_derive_vars) {
print(sprintf("Creating new feature: %s...", new_feat))
args_lst <- NULL
for (arg in glb_derive_lst[[new_feat]]$args)
args_lst[[arg]] <- glb_allobs_df[, arg]
glb_allobs_df[, new_feat] <- do.call(glb_derive_lst[[new_feat]]$mapfn, args_lst)
}## [1] "Creating new feature: descr.my..."
## [1] "Creating new feature: prdl.descr.my.fctr..."#stop(here")
#hex_vctr <- c("\n", "\211", "\235", "\317", "\333")
hex_regex <- paste0(c("\n", "\211", "\235", "\317", "\333"), collapse="|")
for (obs_id in c(10029, 10948, 10136, 10178, 11514, 11904, 12157, 12210, 12659)) {
# tmp_str <- unlist(strsplit(glb_allobs_df[row_pos, "descr.my"], ""))
# glb_allobs_df[row_pos, "descr.my"] <- paste0(tmp_str[!tmp_str %in% hex_vctr],
# collapse="")
row_pos <- which(glb_allobs_df$UniqueID == obs_id)
# glb_allobs_df[row_pos, "descr.my"] <-
# gsub(hex_regex, " ", glb_allobs_df[row_pos, "descr.my"])
}Step 2.2: transform data
#```{r extract_features, cache=FALSE, eval=!is.null(glb_txt_vars)}
glb_chunks_df <- myadd_chunk(glb_chunks_df, "extract.features", major.inc=TRUE)## label step_major step_minor label_minor bgn end elapsed
## 4 transform.data 2 2 2 27.673 27.765 0.092
## 5 extract.features 3 0 0 27.765 NA NAextract.features_chunk_df <- myadd_chunk(NULL, "extract.features_bgn")## label step_major step_minor label_minor bgn end
## 1 extract.features_bgn 1 0 0 27.773 NA
## elapsed
## 1 NA# Create new features that help prediction
# <col_name>.lag.2 <- lag(zoo(glb_trnobs_df$<col_name>), -2, na.pad=TRUE)
# glb_trnobs_df[, "<col_name>.lag.2"] <- coredata(<col_name>.lag.2)
# <col_name>.lag.2 <- lag(zoo(glb_newobs_df$<col_name>), -2, na.pad=TRUE)
# glb_newobs_df[, "<col_name>.lag.2"] <- coredata(<col_name>.lag.2)
#
# glb_newobs_df[1, "<col_name>.lag.2"] <- glb_trnobs_df[nrow(glb_trnobs_df) - 1,
# "<col_name>"]
# glb_newobs_df[2, "<col_name>.lag.2"] <- glb_trnobs_df[nrow(glb_trnobs_df),
# "<col_name>"]
# glb_allobs_df <- mutate(glb_allobs_df,
# A.P.http=ifelse(grepl("http",Added,fixed=TRUE), 1, 0)
# )
#
# glb_trnobs_df <- mutate(glb_trnobs_df,
# )
#
# glb_newobs_df <- mutate(glb_newobs_df,
# )
# Convert dates to numbers
# typically, dates come in as chars;
# so this must be done before converting chars to factors
#stop(here"); sav_allobs_df <- glb_allobs_df #; glb_allobs_df <- sav_allobs_df
if (!is.null(glb_date_vars)) {
glb_allobs_df <- cbind(glb_allobs_df,
myextract_dates_df(df=glb_allobs_df, vars=glb_date_vars,
id_vars=glb_id_var, rsp_var=glb_rsp_var))
for (sfx in c("", ".POSIX"))
glb_exclude_vars_as_features <-
union(glb_exclude_vars_as_features,
paste(glb_date_vars, sfx, sep=""))
for (feat in glb_date_vars) {
glb_allobs_df <- orderBy(reformulate(paste0(feat, ".POSIX")), glb_allobs_df)
# print(myplot_scatter(glb_allobs_df, xcol_name=paste0(feat, ".POSIX"),
# ycol_name=glb_rsp_var, colorcol_name=glb_rsp_var))
print(myplot_scatter(glb_allobs_df[glb_allobs_df[, paste0(feat, ".POSIX")] >=
strptime("2012-12-01", "%Y-%m-%d"), ],
xcol_name=paste0(feat, ".POSIX"),
ycol_name=glb_rsp_var, colorcol_name=paste0(feat, ".wkend")))
# Create features that measure the gap between previous timestamp in the data
require(zoo)
z <- zoo(as.numeric(as.POSIXlt(glb_allobs_df[, paste0(feat, ".POSIX")])))
glb_allobs_df[, paste0(feat, ".zoo")] <- z
print(head(glb_allobs_df[, c(glb_id_var, feat, paste0(feat, ".zoo"))]))
print(myplot_scatter(glb_allobs_df[glb_allobs_df[, paste0(feat, ".POSIX")] >
strptime("2012-10-01", "%Y-%m-%d"), ],
xcol_name=paste0(feat, ".zoo"), ycol_name=glb_rsp_var,
colorcol_name=glb_rsp_var))
b <- zoo(, seq(nrow(glb_allobs_df)))
last1 <- as.numeric(merge(z-lag(z, -1), b, all=TRUE)); last1[is.na(last1)] <- 0
glb_allobs_df[, paste0(feat, ".last1.log")] <- log(1 + last1)
print(gp <- myplot_box(df=glb_allobs_df[glb_allobs_df[,
paste0(feat, ".last1.log")] > 0, ],
ycol_names=paste0(feat, ".last1.log"),
xcol_name=glb_rsp_var))
last2 <- as.numeric(merge(z-lag(z, -2), b, all=TRUE)); last2[is.na(last2)] <- 0
glb_allobs_df[, paste0(feat, ".last2.log")] <- log(1 + last2)
print(gp <- myplot_box(df=glb_allobs_df[glb_allobs_df[,
paste0(feat, ".last2.log")] > 0, ],
ycol_names=paste0(feat, ".last2.log"),
xcol_name=glb_rsp_var))
last10 <- as.numeric(merge(z-lag(z, -10), b, all=TRUE)); last10[is.na(last10)] <- 0
glb_allobs_df[, paste0(feat, ".last10.log")] <- log(1 + last10)
print(gp <- myplot_box(df=glb_allobs_df[glb_allobs_df[,
paste0(feat, ".last10.log")] > 0, ],
ycol_names=paste0(feat, ".last10.log"),
xcol_name=glb_rsp_var))
last100 <- as.numeric(merge(z-lag(z, -100), b, all=TRUE)); last100[is.na(last100)] <- 0
glb_allobs_df[, paste0(feat, ".last100.log")] <- log(1 + last100)
print(gp <- myplot_box(df=glb_allobs_df[glb_allobs_df[,
paste0(feat, ".last100.log")] > 0, ],
ycol_names=paste0(feat, ".last100.log"),
xcol_name=glb_rsp_var))
glb_allobs_df <- orderBy(reformulate(glb_id_var), glb_allobs_df)
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features,
c(paste0(feat, ".zoo")))
# all2$last3 = as.numeric(merge(z-lag(z, -3), b, all = TRUE))
# all2$last5 = as.numeric(merge(z-lag(z, -5), b, all = TRUE))
# all2$last10 = as.numeric(merge(z-lag(z, -10), b, all = TRUE))
# all2$last20 = as.numeric(merge(z-lag(z, -20), b, all = TRUE))
# all2$last50 = as.numeric(merge(z-lag(z, -50), b, all = TRUE))
#
#
# # order table
# all2 = all2[order(all2$id),]
#
# ## fill in NAs
# # count averages
# na.avg = all2 %>% group_by(weekend, hour) %>% dplyr::summarise(
# last1=mean(last1, na.rm=TRUE),
# last3=mean(last3, na.rm=TRUE),
# last5=mean(last5, na.rm=TRUE),
# last10=mean(last10, na.rm=TRUE),
# last20=mean(last20, na.rm=TRUE),
# last50=mean(last50, na.rm=TRUE)
# )
#
# # fill in averages
# na.merge = merge(all2, na.avg, by=c("weekend","hour"))
# na.merge = na.merge[order(na.merge$id),]
# for(i in c("last1", "last3", "last5", "last10", "last20", "last50")) {
# y = paste0(i, ".y")
# idx = is.na(all2[[i]])
# all2[idx,][[i]] <- na.merge[idx,][[y]]
# }
# rm(na.avg, na.merge, b, i, idx, n, pd, sec, sh, y, z)
}
}
rm(last1, last10, last100)## Warning in rm(last1, last10, last100): object 'last1' not found## Warning in rm(last1, last10, last100): object 'last10' not found## Warning in rm(last1, last10, last100): object 'last100' not found# Create factors of string variables
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "factorize.str.vars"), major.inc=TRUE)## label step_major step_minor label_minor
## 1 extract.features_bgn 1 0 0
## 2 extract.features_factorize.str.vars 2 0 0
## bgn end elapsed
## 1 27.773 27.788 0.015
## 2 27.789 NA NA#stop(here"); sav_allobs_df <- glb_allobs_df; #glb_allobs_df <- sav_allobs_df
print(str_vars <- myfind_chr_cols_df(glb_allobs_df))## description condition cellular carrier color
## "description" "condition" "cellular" "carrier" "color"
## storage productline .src .grpid descr.my
## "storage" "productline" ".src" ".grpid" "descr.my"if (length(str_vars <- setdiff(str_vars,
c(glb_exclude_vars_as_features, glb_txt_vars))) > 0) {
for (var in str_vars) {
warning("Creating factors of string variable: ", var,
": # of unique values: ", length(unique(glb_allobs_df[, var])))
glb_allobs_df[, paste0(var, ".fctr")] <-
relevel(factor(glb_allobs_df[, var]),
names(which.max(table(glb_allobs_df[, var], useNA = "ifany"))))
}
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features, str_vars)
}## Warning: Creating factors of string variable: condition: # of unique
## values: 6## Warning: Creating factors of string variable: cellular: # of unique values:
## 3## Warning: Creating factors of string variable: carrier: # of unique values:
## 7## Warning: Creating factors of string variable: color: # of unique values: 5## Warning: Creating factors of string variable: storage: # of unique values:
## 5if (!is.null(glb_txt_vars)) {
require(foreach)
require(gsubfn)
require(stringr)
require(tm)
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "process.text"), major.inc=TRUE)
chk_pattern_freq <- function(rex_str, ignore.case=TRUE) {
match_mtrx <- str_extract_all(txt_vctr, regex(rex_str, ignore_case=ignore.case),
simplify=TRUE)
match_df <- as.data.frame(match_mtrx[match_mtrx != ""])
names(match_df) <- "pattern"
return(mycreate_sqlxtab_df(match_df, "pattern"))
}
# match_lst <- gregexpr("\\bok(?!ay)", txt_vctr[746], ignore.case = FALSE, perl=TRUE); print(match_lst)
dsp_pattern <- function(rex_str, ignore.case=TRUE, print.all=TRUE) {
match_lst <- gregexpr(rex_str, txt_vctr, ignore.case = ignore.case, perl=TRUE)
match_lst <- regmatches(txt_vctr, match_lst)
match_df <- data.frame(matches=sapply(match_lst,
function (elems) paste(elems, collapse="#")))
match_df <- subset(match_df, matches != "")
if (print.all)
print(match_df)
return(match_df)
}
dsp_matches <- function(rex_str, ix) {
print(match_pos <- gregexpr(rex_str, txt_vctr[ix], perl=TRUE))
print(str_sub(txt_vctr[ix], (match_pos[[1]] / 100) * 99 + 0,
(match_pos[[1]] / 100) * 100 + 100))
}
myapply_gsub <- function(...) {
if ((length_lst <- length(names(gsub_map_lst))) == 0)
return(txt_vctr)
for (ptn_ix in 1:length_lst) {
if ((ptn_ix %% 10) == 0)
print(sprintf("running gsub for %02d (of %02d): #%s#...", ptn_ix,
length(names(gsub_map_lst)), names(gsub_map_lst)[ptn_ix]))
txt_vctr <- gsub(names(gsub_map_lst)[ptn_ix], gsub_map_lst[[ptn_ix]],
txt_vctr, ...)
}
return(txt_vctr)
}
myapply_txtmap <- function(txt_vctr, ...) {
nrows <- nrow(glb_txt_map_df)
for (ptn_ix in 1:nrows) {
if ((ptn_ix %% 10) == 0)
print(sprintf("running gsub for %02d (of %02d): #%s#...", ptn_ix,
nrows, glb_txt_map_df[ptn_ix, "rex_str"]))
txt_vctr <- gsub(glb_txt_map_df[ptn_ix, "rex_str"],
glb_txt_map_df[ptn_ix, "rpl_str"],
txt_vctr, ...)
}
return(txt_vctr)
#print(txt_vctr <- glb_allobs_df[glb_allobs_df$UniqueID == 11329, "descr.my"])
#strsplit(txt_vctr, "")[[1]][1]
#ptn_ix <- 2; glb_txt_map_df[ptn_ix, ]
#gsub(glb_txt_map_df[ptn_ix, "rex_str"], glb_txt_map_df[ptn_ix, "rpl_str"], txt_vctr)
#print(match_lst <- gregexpr(glb_txt_map_df[ptn_ix, "rex_str"], txt_vctr))
#strsplit(glb_txt_map_df[ptn_ix, "rex_str"], "")[[1]]
}
chk.equal <- function(bgn, end) {
print(all.equal(sav_txt_lst[["Headline"]][bgn:end],
glb_txt_chr_lst[["Headline"]][bgn:end]))
}
dsp.equal <- function(bgn, end) {
print(sav_txt_lst[["Headline"]][bgn:end])
print(glb_txt_chr_lst[["Headline"]][bgn:end])
}
#sav_txt_lst <- glb_txt_chr_lst; all.equal(sav_txt_lst, glb_txt_chr_lst)
#all.equal(sav_txt_lst[["Headline"]][1:4200], glb_txt_chr_lst[["Headline"]][1:4200])
#chk.equal( 1, 100)
#dsp.equal(86, 90)
#stop(here"); glb_to_sav(); glb_allobs_df <- sav_allobs_df
txt_map_filename <- paste0(glb_txt_munge_filenames_pfx, "map.csv")
if (!file.exists(txt_map_filename))
stop(txt_map_filename, " not found!")
glb_txt_map_df <- read.csv(txt_map_filename, comment.char="#", strip.white=TRUE)
glb_txt_chr_lst <- list();
print(sprintf("Building glb_txt_chr_lst..."))
glb_txt_chr_lst <- foreach(txt_var=glb_txt_vars) %dopar% {
# for (txt_var in glb_txt_vars) {
txt_vctr <- glb_allobs_df[, txt_var]
names(txt_vctr) <- glb_allobs_df[, glb_id_var]
# myapply_txtmap shd be created as a tm_map::content_transformer ?
#print(glb_txt_map_df)
#txt_var=glb_txt_vars[3]; txt_vctr <- glb_txt_chr_lst[[txt_var]]
#print(rex_str <- glb_txt_map_df[3, "rex_str"])
#print(rex_str <- glb_txt_map_df[glb_txt_map_df$rex_str == "\\bWall St\\.", "rex_str"])
#print(rex_str <- glb_txt_map_df[grepl("du Pont", glb_txt_map_df$rex_str), "rex_str"])
#print(rex_str <- glb_txt_map_df[glb_txt_map_df$rpl_str == "versus", "rex_str"])
#print(tmp_vctr <- grep(rex_str, txt_vctr, value=TRUE, ignore.case=FALSE))
#ret_lst <- regexec(rex_str, txt_vctr, ignore.case=FALSE); ret_lst <- regmatches(txt_vctr, ret_lst); ret_vctr <- sapply(1:length(ret_lst), function(pos_ix) ifelse(length(ret_lst[[pos_ix]]) > 0, ret_lst[[pos_ix]], "")); print(ret_vctr <- ret_vctr[ret_vctr != ""])
#gsub(rex_str, glb_txt_map_df[glb_txt_map_df$rex_str == rex_str, "rpl_str"], tmp_vctr, ignore.case=FALSE)
#grep("Hong Hong", txt_vctr, value=TRUE)
txt_vctr <- myapply_txtmap(txt_vctr, ignore.case=FALSE)
}
names(glb_txt_chr_lst) <- glb_txt_vars
for (txt_var in glb_txt_vars) {
print(sprintf("Remaining OK in %s:", txt_var))
txt_vctr <- glb_txt_chr_lst[[txt_var]]
print(chk_pattern_freq(rex_str <- "(?<!(BO|HO|LO))OK(?!(E\\!|ED|IE|IN|S ))",
ignore.case=FALSE))
match_df <- dsp_pattern(rex_str, ignore.case=FALSE, print.all=FALSE)
for (row in row.names(match_df))
dsp_matches(rex_str, ix=as.numeric(row))
print(chk_pattern_freq(rex_str <- "Ok(?!(a\\.|ay|in|ra|um))", ignore.case=FALSE))
match_df <- dsp_pattern(rex_str, ignore.case=FALSE, print.all=FALSE)
for (row in row.names(match_df))
dsp_matches(rex_str, ix=as.numeric(row))
print(chk_pattern_freq(rex_str <- "(?<!( b| B| c| C| g| G| j| M| p| P| w| W| r| Z|\\(b|ar|bo|Bo|co|Co|Ew|gk|go|ho|ig|jo|kb|ke|Ke|ki|lo|Lo|mo|mt|no|No|po|ra|ro|sm|Sm|Sp|to|To))ok(?!(ay|bo|e |e\\)|e,|e\\.|eb|ed|el|en|er|es|ey|i |ie|in|it|ka|ke|ki|ly|on|oy|ra|st|u |uc|uy|yl|yo))",
ignore.case=FALSE))
match_df <- dsp_pattern(rex_str, ignore.case=FALSE, print.all=FALSE)
for (row in row.names(match_df))
dsp_matches(rex_str, ix=as.numeric(row))
}
# txt_vctr <- glb_txt_chr_lst[[glb_txt_vars[1]]]
# print(chk_pattern_freq(rex_str <- "(?<!( b| c| C| p|\\(b|bo|co|lo|Lo|Sp|to|To))ok(?!(ay|e |e\\)|e,|e\\.|ed|el|en|es|ey|ie|in|on|ra))", ignore.case=FALSE))
# print(chk_pattern_freq(rex_str <- "ok(?!(ay|el|on|ra))", ignore.case=FALSE))
# dsp_pattern(rex_str, ignore.case=FALSE, print.all=FALSE)
# dsp_matches(rex_str, ix=8)
# substr(txt_vctr[86], 5613, 5620)
# substr(glb_allobs_df[301, "review"], 550, 650)
#stop(here"); sav_txt_lst <- glb_txt_chr_lst
for (txt_var in glb_txt_vars) {
print(sprintf("Remaining Acronyms in %s:", txt_var))
txt_vctr <- glb_txt_chr_lst[[txt_var]]
print(chk_pattern_freq(rex_str <- "([[:upper:]]\\.( *)){2,}", ignore.case=FALSE))
# Check for names
print(subset(chk_pattern_freq(rex_str <- "(([[:upper:]]+)\\.( *)){1}",
ignore.case=FALSE),
.n > 1))
# dsp_pattern(rex_str="(OK\\.( *)){1}", ignore.case=FALSE)
# dsp_matches(rex_str="(OK\\.( *)){1}", ix=557)
#dsp_matches(rex_str="\\bR\\.I\\.P(\\.*)(\\B)", ix=461)
#dsp_matches(rex_str="\\bR\\.I\\.P(\\.*)", ix=461)
#print(str_sub(txt_vctr[676], 10100, 10200))
#print(str_sub(txt_vctr[74], 1, -1))
}
for (txt_var in glb_txt_vars) {
re_str <- "\\b(Fort|Ft\\.|Hong|Las|Los|New|Puerto|Saint|San|St\\.)( |-)(\\w)+"
print(sprintf("Remaining #%s# terms in %s: ", re_str, txt_var))
txt_vctr <- glb_txt_chr_lst[[txt_var]]
print(orderBy(~ -.n +pattern, subset(chk_pattern_freq(re_str, ignore.case=FALSE),
grepl("( |-)[[:upper:]]", pattern))))
print(" consider cleaning if relevant to problem domain; geography name; .n > 1")
#grep("New G", txt_vctr, value=TRUE, ignore.case=FALSE)
#grep("St\\. Wins", txt_vctr, value=TRUE, ignore.case=FALSE)
}
#stop(here"); sav_txt_lst <- glb_txt_chr_lst
for (txt_var in glb_txt_vars) {
re_str <- "\\b(N|S|E|W|C)( |\\.)(\\w)+"
print(sprintf("Remaining #%s# terms in %s: ", re_str, txt_var))
txt_vctr <- glb_txt_chr_lst[[txt_var]]
print(orderBy(~ -.n +pattern, subset(chk_pattern_freq(re_str, ignore.case=FALSE),
grepl(".", pattern))))
#grep("N Weaver", txt_vctr, value=TRUE, ignore.case=FALSE)
}
for (txt_var in glb_txt_vars) {
re_str <- "\\b(North|South|East|West|Central)( |\\.)(\\w)+"
print(sprintf("Remaining #%s# terms in %s: ", re_str, txt_var))
txt_vctr <- glb_txt_chr_lst[[txt_var]]
if (nrow(filtered_df <- subset(chk_pattern_freq(re_str, ignore.case=FALSE),
grepl(".", pattern))) > 0)
print(orderBy(~ -.n +pattern, filtered_df))
#grep("Central (African|Bankers|Cast|Italy|Role|Spring)", txt_vctr, value=TRUE, ignore.case=FALSE)
#grep("East (Africa|Berlin|London|Poland|Rivals|Spring)", txt_vctr, value=TRUE, ignore.case=FALSE)
#grep("North (American|Korean|West)", txt_vctr, value=TRUE, ignore.case=FALSE)
#grep("South (Pacific|Street)", txt_vctr, value=TRUE, ignore.case=FALSE)
#grep("St\\. Martins", txt_vctr, value=TRUE, ignore.case=FALSE)
}
find_cmpnd_wrds <- function(txt_vctr) {
# Enhancements:
# - arg should be txt_corpus instead of txt_vctr
txt_corpus <- Corpus(VectorSource(txt_vctr))
txt_corpus <- tm_map(txt_corpus, content_transformer(tolower), lazy=TRUE)
txt_corpus <- tm_map(txt_corpus, PlainTextDocument, lazy=TRUE)
txt_corpus <- tm_map(txt_corpus, removePunctuation,
preserve_intra_word_dashes=TRUE, lazy=FALSE)
# Defaulting to Tf since TfIdf with normalize = TRUE throws a warning for empty docs
terms_mtrx <- as.matrix(TermDocumentMatrix(txt_corpus, control=list(weighting=weightTf)))
terms_df <- orderBy(~ -Tf, data.frame(term=dimnames(terms_mtrx)$Terms,
Tf=rowSums(terms_mtrx)))
cmpnd_df <- subset(terms_df, grepl("-", term))
if (nrow(cmpnd_df) == 0) {
print(" No compounded terms found")
return(FALSE)
}
txt_compound_filename <- paste0(glb_txt_munge_filenames_pfx, "compound.csv")
if (!file.exists(txt_compound_filename))
stop(txt_compound_filename, " not found!")
filter_df <- read.csv(txt_compound_filename, comment.char="#", strip.white=TRUE)
cmpnd_df$filter <- FALSE
for (row_ix in 1:nrow(filter_df))
cmpnd_df[!cmpnd_df$filter, "filter"] <-
grepl(filter_df[row_ix, "rex_str"],
cmpnd_df[!cmpnd_df$filter, "term"], ignore.case=TRUE)
cmpnd_df <- subset(cmpnd_df, !filter)
# Bug in tm_map(txt_corpus, removePunctuation, preserve_intra_word_dashes=TRUE) ???
# "net-a-porter" gets converted to "net-aporter"
#grep("net-a-porter", txt_vctr, ignore.case=TRUE, value=TRUE)
#grep("maser-laser", txt_vctr, ignore.case=TRUE, value=TRUE)
#txt_corpus[[which(grepl("net-a-porter", txt_vctr, ignore.case=TRUE))]]
#grep("\\b(across|longer)-(\\w)", cmpnd_df$term, ignore.case=TRUE, value=TRUE)
#grep("(\\w)-(affected|term)\\b", cmpnd_df$term, ignore.case=TRUE, value=TRUE)
print(sprintf("nrow(cmpnd_df): %d", nrow(cmpnd_df)))
myprint_df(cmpnd_df)
}
# This should be run after glb_txt_corpus_lst is created with tolower
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "process.text_reporting_compound_terms"), major.inc=FALSE)
for (txt_var in glb_txt_vars) {
print(sprintf("Remaining compound terms in %s: ", txt_var))
find_cmpnd_wrds(txt_vctr=glb_txt_chr_lst[[txt_var]])
#grep("thirty-five", txt_vctr, ignore.case=TRUE, value=TRUE)
#rex_str <- glb_txt_map_df[grepl("hirty", glb_txt_map_df$rex_str), "rex_str"]
}
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "build.corpus"), major.inc=TRUE)
get_txt_terms <- function(terms_TDM) {
terms_mtrx <- as.matrix(as.TermDocumentMatrix(terms_TDM))
terms_df <- data.frame(term=dimnames(terms_mtrx)$Terms, weight=rowSums(terms_mtrx),
freq=rowSums(terms_mtrx > 0))
terms_df$pos <- 1:nrow(terms_df)
terms_df$cor.y <-
cor(as.matrix(as.DocumentTermMatrix(terms_TDM))[glb_allobs_df$.src == "Train",],
as.numeric(glb_allobs_df[glb_allobs_df$.src == "Train", glb_rsp_var]),
use="pairwise.complete.obs")
terms_df$cor.y.abs <- abs(terms_df$cor.y)
for (cls in unique(glb_allobs_df[, glb_txt_cor_var])) {
if (!is.na(cls))
terms_df[, paste0("weight.", as.character(cls))] <-
colSums(t(terms_mtrx) *
as.numeric(!is.na(glb_allobs_df[, glb_txt_cor_var]) &
(glb_allobs_df[, glb_txt_cor_var] == cls))) else
terms_df[, paste0("weight.", as.character(cls))] <-
colSums(t(terms_mtrx) *
as.numeric(is.na(glb_allobs_df[, glb_txt_cor_var])))
}
# Check all calls to get_terms_DTM_terms to change returned order assumption
return(terms_df <- orderBy(~ -weight, terms_df))
}
#plt_full_df <- get_terms_DTM_terms(terms_DTM=glb_full_terms_DTM_lst[[txt_var]])
get_corpus_terms <- function(txt_corpus) {
return(terms_df <- get_txt_terms(terms_TDM=TermDocumentMatrix(txt_corpus,
control=glb_txt_terms_control)))
}
#stop(here"); glb_to_sav()
glb_txt_corpus_lst <- list()
print(sprintf("Building glb_txt_corpus_lst..."))
glb_txt_corpus_lst <- foreach(txt_var=glb_txt_vars) %dopar% {
# for (txt_var in glb_txt_vars) {
txt_corpus <- Corpus(VectorSource(glb_txt_chr_lst[[txt_var]]))
txt_corpus <- tm_map(txt_corpus, PlainTextDocument, lazy=FALSE)
txt_corpus <- tm_map(txt_corpus, content_transformer(tolower), lazy=FALSE) #nuppr
# removePunctuation does not replace with whitespace. Use a custom transformer ???
txt_corpus <- tm_map(txt_corpus, removePunctuation, lazy=TRUE) #npnct<chr_ix>
# txt-corpus <- tm_map(txt_corpus, content_transformer(function(x, pattern) gsub(pattern, "", x))
if (!is.null(glb_txt_stop_words[[txt_var]]))
txt_corpus <- tm_map(txt_corpus, removeWords, glb_txt_stop_words[[txt_var]],
lazy=FALSE)#, lazy=TRUE) #nstopwrds
#print("StoppedWords:"); stopped_words_TfIdf_df <- inspect_terms(txt_corpus)
#stopped_words_TfIdf_df[grepl("cond", stopped_words_TfIdf_df$term, ignore.case=TRUE), ]
#txt_X_mtrx <- as.matrix(DocumentTermMatrix(txt_corpus, control=list(weighting=weightTfIdf)))
#which(txt_X_mtrx[, 211] > 0)
#glb_allobs_df[which(txt_X_mtrx[, 211] > 0), glb_txt_vars]
#txt_X_mtrx[2159, txt_X_mtrx[2159, ] > 0]
# txt_corpus <- tm_map(txt_corpus, stemDocument, "english", lazy=TRUE) #Done below
#txt_corpus <- tm_map(txt_corpus, content_transformer(stemDocument))
#print("StemmedWords:"); stemmed_words_TfIdf_df <- inspect_terms(txt_corpus)
#stemmed_words_TfIdf_df[grepl("cond", stemmed_words_TfIdf_df$term, ignore.case=TRUE), ]
#stm_X_mtrx <- as.matrix(DocumentTermMatrix(txt_corpus, control=list(weighting=weightTfIdf)))
#glb_allobs_df[which((stm_X_mtrx[, 180] > 0) | (stm_X_mtrx[, 181] > 0)), glb_txt_vars]
#glb_allobs_df[which((stm_X_mtrx[, 181] > 0)), glb_txt_vars]
# glb_txt_corpus_lst[[txt_var]] <- txt_corpus
}
names(glb_txt_corpus_lst) <- glb_txt_vars
mycombineSynonyms <- content_transformer(function(x, syn=NULL) {
Reduce(function(a,b) {
gsub(paste0("\\b(", paste(b$syns, collapse="|"),")\\b"), b$word, a)}, syn, x)
})
#stop(here"); glb_to_sav(); sav_txt_corpus <- glb_txt_corpus_lst[[txt_var]]; all.equal(sav_txt_corpus, glb_txt_corpus_lst[[txt_var]]); glb_txt_corpus_lst[[txt_var]] <- sav_txt_corpus
glb_post_stop_words_terms_df_lst <- list();
glb_post_stop_words_terms_mtrx_lst <- list();
glb_post_stem_words_terms_df_lst <- list();
glb_post_stem_words_terms_mtrx_lst <- list();
for (txt_var in glb_txt_vars) {
print(sprintf(" Top_n stop term weights for %s:", txt_var))
# This impacts stemming probably due to lazy parameter
print(myprint_df(full_terms_df <-
get_corpus_terms(txt_corpus=glb_txt_corpus_lst[[txt_var]]),
glb_txt_top_n[[txt_var]]))
glb_post_stop_words_terms_df_lst[[txt_var]] <- full_terms_df
terms_stop_mtrx <- as.matrix(DocumentTermMatrix(glb_txt_corpus_lst[[txt_var]],
control=glb_txt_terms_control))
rownames(terms_stop_mtrx) <- rownames(glb_allobs_df) # print undreadable otherwise
glb_post_stop_words_terms_mtrx_lst[[txt_var]] <- terms_stop_mtrx
tmp_allobs_df <- glb_allobs_df[, c(glb_id_var, glb_rsp_var)]
tmp_allobs_df$terms.post.stop.n <- rowSums(terms_stop_mtrx > 0)
tmp_allobs_df$terms.post.stop.n.log <- log(1 + tmp_allobs_df$terms.post.stop.n)
tmp_allobs_df$weight.post.stop.sum <- rowSums(terms_stop_mtrx)
print(sprintf(" Top_n stem term weights for %s:", txt_var))
glb_txt_corpus_lst[[txt_var]] <- tm_map(glb_txt_corpus_lst[[txt_var]], stemDocument,
"english", lazy=FALSE)
if (!is.null(glb_txt_synonyms[[txt_var]])) {
syn_lst <- myrmNullObj(glb_txt_synonyms[[txt_var]])
glb_txt_corpus_lst[[txt_var]] <- tm_map(glb_txt_corpus_lst[[txt_var]],
mycombineSynonyms,
syn_lst, lazy=FALSE)
}
print(myprint_df(full_terms_df <- get_corpus_terms(glb_txt_corpus_lst[[txt_var]]),
glb_txt_top_n[[txt_var]]))
glb_post_stem_words_terms_df_lst[[txt_var]] <- full_terms_df
terms_stem_mtrx <- as.matrix(DocumentTermMatrix(glb_txt_corpus_lst[[txt_var]],
control=glb_txt_terms_control))
rownames(terms_stem_mtrx) <- rownames(glb_allobs_df) # print undreadable otherwise
glb_post_stem_words_terms_mtrx_lst[[txt_var]] <- terms_stem_mtrx
tmp_allobs_df$terms.post.stem.n <- rowSums(terms_stem_mtrx > 0)
tmp_allobs_df$terms.post.stem.n.log <- log(1 + tmp_allobs_df$terms.post.stem.n)
tmp_allobs_df$weight.post.stem.sum <- rowSums(terms_stem_mtrx)
tmp_allobs_df$terms.n.stem.stop.Ratio <-
1.0 * tmp_allobs_df$terms.post.stem.n / tmp_allobs_df$terms.post.stop.n
tmp_allobs_df[(is.nan(tmp_allobs_df$terms.n.stem.stop.Ratio) |
is.infinite(tmp_allobs_df$terms.n.stem.stop.Ratio)),
"terms.n.stem.stop.Ratio"] <- 1.0
if ((n.errors <- sum(tmp_allobs_df$terms.n.stem.stop.Ratio > 1)) > 0)
stop(n.errors, " obs in tmp_allobs_df have terms.n.stem.stop.Ratio > 1",
" happening due to terms filtered by glb_txt_terms_control$bounds$global[1] but stemmable to other terms")
#print(head(subset(tmp_allobs_df, terms.n.stem.stop.Ratio > 1)))
#glb_allobs_df[(row_ix <- which(glb_allobs_df$UniqueID == 10465)), ]
#terms_stop_mtrx[row_ix, terms_stop_mtrx[row_ix, ] > 0]
#setdiff(names(terms_stem_mtrx[row_ix, terms_stem_mtrx[row_ix, ] > 0]), names(terms_stop_mtrx[row_ix, terms_stop_mtrx[row_ix, ] > 0]))
#mydsp_obs(list(descr.my.contains="updat"))
tmp_allobs_df$weight.sum.stem.stop.Ratio <-
1.0 * tmp_allobs_df$weight.post.stem.sum / tmp_allobs_df$weight.post.stop.sum
tmp_allobs_df[is.nan(tmp_allobs_df$weight.sum.stem.stop.Ratio) |
is.infinite(tmp_allobs_df$weight.sum.stem.stop.Ratio),
"weight.sum.stem.stop.Ratio"] <- 1.0
tmp_trnobs_df <- tmp_allobs_df[!is.na(tmp_allobs_df[, glb_rsp_var]), ]
print(cor(as.matrix(tmp_trnobs_df[, -c(1, 2)]),
as.numeric(tmp_trnobs_df[, glb_rsp_var])))
txt_var_pfx <- toupper(substr(txt_var, 1, 1))
tmp_allobs_df <- tmp_allobs_df[, -c(1, 2)]
names(tmp_allobs_df) <- paste(paste0(txt_var_pfx, "."), names(tmp_allobs_df),
sep="")
glb_allobs_df <- cbind(glb_allobs_df, tmp_allobs_df)
glb_exclude_vars_as_features <- c(glb_exclude_vars_as_features,
paste(paste0(txt_var_pfx, ".terms.post."), c("stop.n", "stem.n"), sep=""))
}
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "extract.DTM"), major.inc=TRUE)
#stop(here")
glb_full_DTM_lst <- list(); glb_sprs_DTM_lst <- list();
for (txt_var in glb_txt_vars) {
print(sprintf("Extracting term weights for %s...", txt_var))
txt_corpus <- glb_txt_corpus_lst[[txt_var]]
full_DTM <- DocumentTermMatrix(txt_corpus,
control=glb_txt_terms_control)
sprs_DTM <- removeSparseTerms(full_DTM,
glb_sprs_thresholds[txt_var])
glb_full_DTM_lst[[txt_var]] <- full_DTM
glb_sprs_DTM_lst[[txt_var]] <- sprs_DTM
}
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "report.DTM"), major.inc=TRUE)
require(reshape2)
for (txt_var in glb_txt_vars) {
print(sprintf("Reporting term weights for %s...", txt_var))
full_DTM <- glb_full_DTM_lst[[txt_var]]
sprs_DTM <- glb_sprs_DTM_lst[[txt_var]]
print(" Full TermMatrix:"); print(full_DTM)
full_terms_df <- get_txt_terms(full_DTM)
# full_terms_df <- full_terms_df[, c(2, 1, 3, 4)]
# col_names <- names(full_terms_df)
# col_names[2:length(col_names)] <-
# paste(col_names[2:length(col_names)], ".full", sep="")
# names(full_terms_df) <- col_names
print(" Sparse TermMatrix:"); print(sprs_DTM)
sprs_terms_df <- get_txt_terms(sprs_DTM)
# sprs_terms_df <- sprs_terms_df[, c(2, 1, 3, 4)]
# col_names <- names(sprs_terms_df)
# col_names[2:length(col_names)] <-
# paste(col_names[2:length(col_names)], ".sprs", sep="")
# names(sprs_terms_df) <- col_names
intersect(names(full_terms_df), names(sprs_terms_df))
terms_df <- merge(full_terms_df, sprs_terms_df,
by=setdiff(intersect(names(full_terms_df), names(sprs_terms_df)), "pos"),
all.x=TRUE, suffixes=c(".full", ".sprs"))
terms_df$in.sprs <- !is.na(terms_df$pos.sprs)
plt_terms_df <- subset(terms_df,
weight >= min(terms_df$weight[!is.na(terms_df$pos.sprs)], na.rm=TRUE))
plt_terms_df$label <- ""
plt_terms_df[is.na(plt_terms_df$pos.sprs), "label"] <-
plt_terms_df[is.na(plt_terms_df$pos.sprs), "term"]
# glb_important_terms[[txt_var]] <- union(glb_important_terms[[txt_var]],
# plt_terms_df[is.na(plt_terms_df$TfIdf.sprs), "term"])
print(myplot_scatter(plt_terms_df, "freq", "weight",
colorcol_name="in.sprs") +
geom_text(aes(label=label), color="Black", size=3.5))
melt_terms_df <- orderBy(~ -value,
melt(terms_df, id.vars="term", measure.vars = c("weight", "freq")))
print(ggplot(melt_terms_df, aes(value, color=variable)) + stat_ecdf() +
geom_hline(yintercept=glb_sprs_thresholds[txt_var],
linetype = "dotted"))
melt_terms_df <- orderBy(~ -value,
melt(subset(terms_df, in.sprs), id.vars="term",
measure.vars=grep("weight.", names(terms_df), value=TRUE)))
print(myplot_hbar(melt_terms_df, "term", "value", colorcol_name="variable"))
melt_terms_df <- orderBy(~ -value,
melt(subset(terms_df, !in.sprs), id.vars="term",
measure.vars=grep("weight.", names(terms_df), value=TRUE)))
print(myplot_hbar(head(melt_terms_df, glb_txt_top_n[[txt_var]]), "term", "value",
colorcol_name="variable"))
}
# sav_full_DTM_lst <- glb_full_DTM_lst
# print(identical(sav_glb_txt_corpus_lst, glb_txt_corpus_lst))
# print(all.equal(length(sav_glb_txt_corpus_lst), length(glb_txt_corpus_lst)))
# print(all.equal(names(sav_glb_txt_corpus_lst), names(glb_txt_corpus_lst)))
# print(all.equal(sav_glb_txt_corpus_lst[["Headline"]], glb_txt_corpus_lst[["Headline"]]))
# print(identical(sav_full_DTM_lst, glb_full_DTM_lst))
rm(full_terms_mtrx)
# Create txt features
if ((length(glb_txt_vars) > 1) &&
(length(unique(pfxs <- sapply(glb_txt_vars,
function(txt) toupper(substr(txt, 1, 1))))) < length(glb_txt_vars)))
stop("Prefixes for corpus freq terms not unique: ", pfxs)
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "bind.DTM"),
major.inc=TRUE)
#stop(here"); glb_to_sav(); all.equal(sav_allobs_df, glb_allobs_df); glb_allobs_df <- sav_allobs_df
require(tidyr)
for (txt_var in glb_txt_vars) {
print(sprintf("Binding DTM for %s...", txt_var))
txt_var_pfx <- toupper(substr(txt_var, 1, 1))
txt_full_X_df <- as.data.frame(as.matrix(glb_full_DTM_lst[[txt_var]]))
terms_full_df <- get_txt_terms(glb_full_DTM_lst[[txt_var]])
colnames(txt_full_X_df) <- paste(txt_var_pfx, ".T.",
make.names(colnames(txt_full_X_df)), sep="")
rownames(txt_full_X_df) <- rownames(glb_allobs_df) # warning otherwise
plt_full_df <- terms_full_df
names(plt_full_df)[grepl("weight$", names(plt_full_df))] <- "weight.all"
# gather(plt_full_df[1:5, ], domain, TfIdf, -matches("!(TfIdf)"))
# gather(plt_full_df[1:5, grepl("TfIdf", names(plt_full_df))], domain, TfIdf)
# gather(plt_full_df[1:5, ], domain, TfIdf,
# -names(plt_full_df)[!grepl("TfIdf", names(plt_full_df))])
plt_full_df <- gather(plt_full_df, domain, weight,
-c(term, freq, pos, cor.y, cor.y.abs))
plt_full_df$label <- NA
top_val_terms <- orderBy(~-weight, terms_full_df)$term[1:glb_txt_top_n[[txt_var]]]
plt_full_df[plt_full_df$term %in% top_val_terms, "label"] <-
plt_full_df[plt_full_df$term %in% top_val_terms, "term"]
top_cor_terms <- orderBy(~-cor.y.abs,
terms_full_df)$term[1:glb_txt_top_n[[txt_var]]]
plt_full_df[plt_full_df$term %in% top_cor_terms, "label"] <-
plt_full_df[plt_full_df$term %in% top_cor_terms, "term"]
#plt_full_df$type <- "none"
plt_full_df[plt_full_df$term %in% top_val_terms, "type"] <- "top.weight"
plt_full_df[plt_full_df$term %in% top_cor_terms, "type"] <- "top.cor"
plt_full_df[plt_full_df$term %in% intersect(top_val_terms, top_cor_terms), "type"] <-
"top.both"
cor.y.rnorm <- cor(glb_allobs_df$.rnorm, glb_allobs_df[, glb_rsp_var],
use="pairwise.complete.obs")
print(ggplot(plt_full_df, aes(x=weight, y=cor.y)) + facet_wrap(~ domain) +
geom_point(aes(size=freq), color="grey") +
geom_jitter() +
geom_text(aes(label=label, color=type), size=3.5) +
#geom_hline(yintercept=cor.y.rnorm, color="red") +
geom_hline(yintercept=c(cor.y.rnorm, -cor.y.rnorm), color="red"))
if (glb_txt_terms_filter == "sparse") {
txt_X_df <- as.data.frame(as.matrix(glb_sprs_DTM_lst[[txt_var]]))
select_terms <- make.names(colnames(txt_X_df))
# colnames(txt_X_df) <- paste(txt_var_pfx, ".T.",
# make.names(colnames(txt_X_df)), sep="")
# rownames(txt_X_df) <- rownames(glb_allobs_df) # warning otherwise
} else if (glb_txt_terms_filter == "top.val") {
select_terms <- orderBy(~-weight,
terms_full_df)$term[1:glb_txt_top_n[[txt_var]]]
# txt_X_df <- txt_full_X_df[, subset(terms_full_df, term %in% select_terms)$pos,
# FALSE]
} else if (glb_txt_terms_filter == "top.cor") {
select_terms <- orderBy(~-cor.y.abs,
terms_full_df)$term[1:glb_txt_top_n[[txt_var]]]
# txt_X_df <- txt_full_X_df[, subset(terms_full_df, term %in% select_terms)$pos,
# FALSE]
} else if (glb_txt_terms_filter == "union.top.val.cor") {
select_terms <- union(
orderBy(~-weight , terms_full_df)$term[1:glb_txt_top_n[[txt_var]]],
orderBy(~-cor.y.abs, terms_full_df)$term[1:glb_txt_top_n[[txt_var]]])
} else stop(
"glb_txt_terms_filter should be one of c('sparse', 'top.val', 'top.cor', 'union.top.val.cor') vs. '",
glb_txt_terms_filter, "'")
assoc_terms_lst <- findAssocs(glb_full_DTM_lst[[txt_var]], select_terms,
glb_txt_assoc_cor[[txt_var]])
assoc_terms <- c(NULL)
for (term in names(assoc_terms_lst))
if (length(assoc_terms_lst[[term]]) > 0)
assoc_terms <- union(assoc_terms, names(assoc_terms_lst[[term]]))
txt_X_df <- txt_full_X_df[,
subset(terms_full_df, term %in% c(select_terms, assoc_terms))$pos,
FALSE]
glb_allobs_df <- cbind(glb_allobs_df, txt_X_df) # TfIdf is normalized
#glb_allobs_df <- cbind(glb_allobs_df, log_X_df) # if using non-normalized metrics
}
#identical(chk_entity_df, glb_allobs_df)
#chk_entity_df <- glb_allobs_df
extract.features_chunk_df <- myadd_chunk(extract.features_chunk_df,
paste0("extract.features_", "bind.DXM"),
major.inc=TRUE)
#stop(here"); sav_allobs_df <- glb_allobs_df; glb_allobs_df <- sav_allobs_df
glb_punct_vctr <- c("!", "\"", "#", "\\$", "%", "&", "'",
"\\(|\\)",# "\\(", "\\)",
"\\*", "\\+", ",", "-", "\\.", "/", ":", ";",
"<|>", # "<",
"=",
# ">",
"\\?", "@", "\\[", "\\\\", "\\]", "\\^", "_", "`",
"\\{", "\\|", "\\}", "~")
txt_X_df <- glb_allobs_df[, c(glb_id_var, ".rnorm"), FALSE]
txt_X_df <- foreach(txt_var=glb_txt_vars, .combine=cbind) %dopar% {
#for (txt_var in glb_txt_vars) {
print(sprintf("Binding DXM for %s...", txt_var))
txt_var_pfx <- toupper(substr(txt_var, 1, 1))
txt_full_DTM_mtrx <- as.matrix(glb_full_DTM_lst[[txt_var]])
rownames(txt_full_DTM_mtrx) <- rownames(glb_allobs_df) # print undreadable otherwise
#print(txt_full_DTM_mtrx[txt_full_DTM_mtrx[, "ebola"] != 0, "ebola"])
# Create <txt_var>.T.<term> for glb_important_terms
for (term in glb_important_terms[[txt_var]])
txt_X_df[, paste0(txt_var_pfx, ".T.", make.names(term))] <-
txt_full_DTM_mtrx[, term]
# Create <txt_var>.wrds.n.log & .wrds.unq.n.log
txt_X_df[, paste0(txt_var_pfx, ".wrds.n.log")] <-
log(1 + mycount_pattern_occ("\\w+", glb_txt_chr_lst[[txt_var]]))
txt_X_df[, paste0(txt_var_pfx, ".wrds.unq.n.log")] <-
log(1 + rowSums(txt_full_DTM_mtrx != 0))
txt_X_df[, paste0(txt_var_pfx, ".weight.sum")] <-
rowSums(txt_full_DTM_mtrx)
txt_X_df[, paste0(txt_var_pfx, ".ratio.weight.sum.wrds.n")] <-
txt_X_df[, paste0(txt_var_pfx, ".weight.sum")] /
(exp(txt_X_df[, paste0(txt_var_pfx, ".wrds.n.log")]) - 1)
txt_X_df[is.nan(txt_X_df[, paste0(txt_var_pfx, ".ratio.weight.sum.wrds.n")]),
paste0(txt_var_pfx, ".ratio.weight.sum.wrds.n")] <- 0
# Create <txt_var>.chrs.n.log
txt_X_df[, paste0(txt_var_pfx, ".chrs.n.log")] <-
log(1 + mycount_pattern_occ(".", glb_allobs_df[, txt_var]))
txt_X_df[, paste0(txt_var_pfx, ".chrs.uppr.n.log")] <-
log(1 + mycount_pattern_occ("[[:upper:]]", glb_allobs_df[, txt_var]))
txt_X_df[, paste0(txt_var_pfx, ".dgts.n.log")] <-
log(1 + mycount_pattern_occ("[[:digit:]]", glb_allobs_df[, txt_var]))
# Create <txt_var>.npnct?.log
# would this be faster if it's iterated over each row instead of
# each created column ???
for (punct_ix in 1:length(glb_punct_vctr)) {
# smp0 <- " "
# smp1 <- "! \" # $ % & ' ( ) * + , - . / : ; < = > ? @ [ \ ] ^ _ ` { | } ~"
# smp2 <- paste(smp1, smp1, sep=" ")
# print(sprintf("Testing %s pattern:", glb_punct_vctr[punct_ix]))
# results <- mycount_pattern_occ(glb_punct_vctr[punct_ix], c(smp0, smp1, smp2))
# names(results) <- NULL; print(results)
txt_X_df[,
paste0(txt_var_pfx, ".chrs.pnct", sprintf("%02d", punct_ix), ".n.log")] <-
log(1 + mycount_pattern_occ(glb_punct_vctr[punct_ix],
glb_allobs_df[, txt_var]))
}
# print(head(glb_allobs_df[glb_allobs_df[, "A.npnct23.log"] > 0,
# c("UniqueID", "Popular", "Abstract", "A.npnct23.log")]))
# Create <txt_var>.wrds.stop.n.log & <txt_var>ratio.wrds.stop.n.wrds.n
if (!is.null(glb_txt_stop_words[[txt_var]])) {
stop_words_rex_str <- paste0("\\b(",
paste0(glb_txt_stop_words[[txt_var]], collapse="|"),
")\\b")
txt_X_df[, paste0(txt_var_pfx, ".wrds.stop.n", ".log")] <-
log(1 + mycount_pattern_occ(stop_words_rex_str, glb_txt_chr_lst[[txt_var]]))
txt_X_df[, paste0(txt_var_pfx, ".ratio.wrds.stop.n.wrds.n")] <-
exp(txt_X_df[, paste0(txt_var_pfx, ".wrds.stop.n", ".log")] -
txt_X_df[, paste0(txt_var_pfx, ".wrds.n", ".log")])
}
# Create <txt_var>.P.http
txt_X_df[, paste(txt_var_pfx, ".P.http", sep="")] <-
as.integer(0 + mycount_pattern_occ("http", glb_allobs_df[, txt_var]))
# Create <txt_var>.P.mini & air
txt_X_df[, paste(txt_var_pfx, ".P.mini", sep="")] <-
as.integer(0 + mycount_pattern_occ("mini(?!m)", glb_allobs_df[, txt_var],
perl=TRUE))
txt_X_df[, paste(txt_var_pfx, ".P.air", sep="")] <-
as.integer(0 + mycount_pattern_occ("(?<![fhp])air", glb_allobs_df[, txt_var],
perl=TRUE))
txt_X_df[, paste(txt_var_pfx, ".P.black", sep="")] <-
as.integer(0 + mycount_pattern_occ("black", glb_allobs_df[, txt_var],
perl=TRUE))
txt_X_df[, paste(txt_var_pfx, ".P.white", sep="")] <-
as.integer(0 + mycount_pattern_occ("white", glb_allobs_df[, txt_var],
perl=TRUE))
txt_X_df[, paste(txt_var_pfx, ".P.gold", sep="")] <-
as.integer(0 + mycount_pattern_occ("gold", glb_allobs_df[, txt_var],
perl=TRUE))
txt_X_df[, paste(txt_var_pfx, ".P.spacegray", sep="")] <-
as.integer(0 + mycount_pattern_occ("spacegray", glb_allobs_df[, txt_var],
perl=TRUE))
txt_X_df <- subset(txt_X_df, select=-.rnorm)
txt_X_df <- txt_X_df[, -grep(glb_id_var, names(txt_X_df), fixed=TRUE), FALSE]
#glb_allobs_df <- cbind(glb_allobs_df, txt_X_df)
}
glb_allobs_df <- cbind(glb_allobs_df, txt_X_df)
#myplot_box(glb_allobs_df, "A.sum.TfIdf", glb_rsp_var)
# if (sum(is.na(glb_allobs_df$D.P.http)) > 0)
# stop("Why is this happening ?")
# Generate summaries
# print(summary(glb_allobs_df))
# print(sapply(names(glb_allobs_df), function(col) sum(is.na(glb_allobs_df[, col]))))
# print(summary(glb_trnobs_df))
# print(sapply(names(glb_trnobs_df), function(col) sum(is.na(glb_trnobs_df[, col]))))
# print(summary(glb_newobs_df))
# print(sapply(names(glb_newobs_df), function(col) sum(is.na(glb_newobs_df[, col]))))
glb_exclude_vars_as_features <- union(glb_exclude_vars_as_features,
glb_txt_vars)
rm(log_X_df, txt_X_df)
}
# Use model info provided in description
# mydsp_obs(list(description.contains="a[[:digit:]]"), cols=glb_dsp_cols, all=TRUE)
# glb_allobs_df[glb_allobs_df$UniqueID == 12474, "prdline.my"] <- "iPad mini"
# glb_allobs_df[glb_allobs_df$UniqueID == 12474, "color"] <- "Space Gray"
# glb_allobs_df[glb_allobs_df$UniqueID == 12474, "cellular"] <- "0"
# glb_allobs_df[glb_allobs_df$UniqueID == 12474, "carrier"] <- "None"
#
# mydsp_obs(list(description.contains="m(.{4})ll"), cols=glb_dsp_cols, all=TRUE)
# glb_allobs_df[glb_allobs_df$UniqueID == 11360, "color"] <- "Black"
# glb_allobs_df[glb_allobs_df$UniqueID == 11360, "storage"] <- "64"
# glb_allobs_df[glb_allobs_df$UniqueID == 11360, "cellular"] <- "0"
# glb_allobs_df[glb_allobs_df$UniqueID == 11360, "carrier"] <- "None"
#
# glb_allobs_df[glb_allobs_df$UniqueID == 11361, "prdline.my"] <- "iPad Air"
# glb_allobs_df[glb_allobs_df$UniqueID == 11361, "storage"] <- "32"
# glb_allobs_df[glb_allobs_df$UniqueID == 11361, "color"] <- "White"
# glb_allobs_df[glb_allobs_df$UniqueID == 11361, "cellular"] <- "0"
# glb_allobs_df[glb_allobs_df$UniqueID == 11361, "carrier"] <- "None"
# mydsp_obs(list(description.contains="mini(?!m)"), perl=TRUE, cols="D.P.mini", all=TRUE)
# mydsp_obs(list(D.P.mini=1), cols="D.P.mini", all=TRUE)
# mydsp_obs(list(D.P.mini=1, productline="Unknown"), cols="D.P.mini", all=TRUE)
# mydsp_obs(list(description.contains="(?<![fhp])air"), perl=TRUE, all=TRUE)
# mydsp_obs(list(description.contains="air"), perl=FALSE, cols="D.P.air", all=TRUE)
# mydsp_obs(list(D.P.air=1, productline="Unknown"), cols="D.P.air", all=TRUE)
# print(mycreate_sqlxtab_df(glb_allobs_df, c("prdline.my", "productline", "D.P.mini",
# glb_rsp_var)))
# print(glb_allobs_df[(glb_allobs_df$productline == "Unknown") &
# (glb_allobs_df$D.P.mini > 0),
# c(glb_id_var, glb_category_var, glb_dsp_cols, glb_txt_vars)])
# glb_allobs_df[(glb_allobs_df$D.P.mini == 1) & (glb_allobs_df$productline == "Unknown"),
# "prdline.my"] <- "iPad mini"
# print(mycreate_sqlxtab_df(glb_allobs_df, c("prdline.my", "productline", "D.P.air",
# glb_rsp_var)))
# print(glb_allobs_df[(glb_allobs_df$productline == "Unknown") &
# (glb_allobs_df$D.P.air > 0),
# c(glb_id_var, glb_category_var, glb_dsp_cols, glb_txt_vars)])
# #glb_allobs_df[glb_allobs_df$UniqueID == 11863, "D.P.air"] <- 0
# glb_allobs_df[(glb_allobs_df$D.P.air == 1) & (glb_allobs_df$productline == "Unknown"),
# "prdline.my"] <- "iPad Air"
# print(glb_allobs_df[(glb_allobs_df$UniqueID %in% c(11767, 11811, 12156)),
# c(glb_id_var, "sold",
# "prdline.my", "color", "condition", "cellular", "carrier", "storage"
# #, "descr.my"
# )])
# glb_allobs_df[glb_allobs_df$UniqueID == 11767, "prdline.my"] <- "iPad 2"
# glb_allobs_df[glb_allobs_df$UniqueID == 11767, "storage"] <- "32"
# glb_allobs_df[glb_allobs_df$UniqueID == 11811, "prdline.my"] <- "iPad 2"
# glb_allobs_df[glb_allobs_df$UniqueID == 12156, "prdline.my"] <- "iPad 1"
# mydsp_obs(list(prdline.my="Unknown"), all=TRUE)
# tmp_allobs_df <- glb_allobs_df[, "prdline.my", FALSE]
# names(tmp_allobs_df) <- "old.prdline.my"
# glb_allobs_df$prdline.my <-
# plyr::revalue(glb_allobs_df$prdline.my, c(
# # "iPad 1" = "iPad",
# # "iPad 2" = "iPad2+",
# "iPad 3" = "iPad 3+",
# "iPad 4" = "iPad 3+",
# "iPad 5" = "iPad 3+",
#
# "iPad Air" = "iPadAir",
# "iPad Air 2" = "iPadAir",
#
# "iPad mini" = "iPadmini",
# "iPad mini 2" = "iPadmini 2+",
# "iPad mini 3" = "iPadmini 2+",
# "iPad mini Retina" = "iPadmini 2+"
# ))
# tmp_allobs_df$prdline.my <- glb_allobs_df[, "prdline.my"]
# print(mycreate_sqlxtab_df(tmp_allobs_df, c("prdline.my", "old.prdline.my")))
# print(mycreate_sqlxtab_df(tmp_allobs_df, c("prdline.my")))
# print(mycreate_sqlxtab_df(subset(glb_allobs_df, color == "Unknown"),
# c("color", "D.P.black", "D.P.gold", "D.P.spacegray", "D.P.white")))
# print(glb_allobs_df[(glb_allobs_df$color == "Unknown") & (glb_allobs_df$D.P.black > 0),
# c(glb_id_var, "color", "D.P.black", "sold", "prdline.my", "condition",
# "cellular", "carrier", "storage", "descr.my")])
# glb_allobs_df[glb_allobs_df$UniqueID == 12137, "color"] <- "Black"
# print(glb_allobs_df[(glb_allobs_df$color == "Unknown") & (glb_allobs_df$D.P.spacegray > 0),
# c(glb_id_var, "color", "D.P.spacegray", "prdline.my", "condition",
# "cellular", "carrier", "storage", "descr.my")])
# glb_allobs_df[glb_allobs_df$UniqueID %in% c(12106), "color"] <- "Space Gray"
# print(glb_allobs_df[(glb_allobs_df$color == "Unknown") & (glb_allobs_df$D.P.white > 0),
# c(glb_id_var, "color", "D.P.white", "prdline.my", "condition",
# "cellular", "carrier", "storage", "descr.my")])
# glb_allobs_df[glb_allobs_df$UniqueID %in% c(10573, 10809, 10925, 11735), "color"] <-
# "White"
glb_allobs_df$carrier.fctr <- as.factor(glb_allobs_df$carrier)
glb_allobs_df$cellular.fctr <- as.factor(glb_allobs_df$cellular)
glb_allobs_df$color.fctr <- as.factor(glb_allobs_df$color)
# glb_allobs_df$prdline.my.fctr <- as.factor(glb_allobs_df$prdline.my)
glb_allobs_df$storage.fctr <- as.factor(glb_allobs_df$storage)
#stop(here"); sav_allobs_df <- glb_allobs_df; glb_allobs_df <- sav_allobs_df
# glb_allobs_df %>%
# unite(prdl.descr.my, c(productline, as.numeric(D.chrs.n.log > 0), sep="#"))
# unite_("prdl.descr.my", interp(~c("productline", as.numeric(D.chrs.n.log > 0), sep="#")))
# print(sapply(names(glb_trnobs_df), function(col) sum(is.na(glb_trnobs_df[, col]))))
# print(sapply(names(glb_newobs_df), function(col) sum(is.na(glb_newobs_df[, col]))))
# print(myplot_scatter(glb_trnobs_df, "<col1_name>", "<col2_name>", smooth=TRUE))
#stop(here"); glb_to_sav(); glb_allobs_df <- sav_allobs_df
if (!is.null(glb_price_vars)) {
for (var in glb_price_vars) {
for (digit in 1:(log10(max(glb_allobs_df[, var], na.rm=TRUE)) + 1)) {
glb_allobs_df[, paste0(var, ".dgt", digit, ".is9")] <-
as.numeric(as.integer((as.integer(glb_allobs_df[, var]) %% (10 ^ digit)) /
(10 ^ (digit - 1))) == 9)
}
for (decimal in 1:2) {
glb_allobs_df[, paste0(var, ".dcm", decimal, ".is9")] <-
as.numeric(as.integer(glb_allobs_df[, var] * (10 ^ decimal)) %% 10 == 9)
}
}
#as.numeric((as.integer(startprice) %% 10) == 9)
}
rm(corpus_lst
, glb_sprs_DTM_lst #, glb_full_DTM_lst
, txt_corpus, txt_vctr)## Warning in rm(corpus_lst, glb_sprs_DTM_lst, txt_corpus, txt_vctr): object
## 'corpus_lst' not found## Warning in rm(corpus_lst, glb_sprs_DTM_lst, txt_corpus, txt_vctr): object
## 'glb_sprs_DTM_lst' not found## Warning in rm(corpus_lst, glb_sprs_DTM_lst, txt_corpus, txt_vctr): object
## 'txt_corpus' not found## Warning in rm(corpus_lst, glb_sprs_DTM_lst, txt_corpus, txt_vctr): object
## 'txt_vctr' not foundextract.features_chunk_df <- myadd_chunk(extract.features_chunk_df, "extract.features_end",
major.inc=TRUE)## label step_major step_minor label_minor
## 2 extract.features_factorize.str.vars 2 0 0
## 3 extract.features_end 3 0 0
## bgn end elapsed
## 2 27.789 27.85 0.061
## 3 27.850 NA NAmyplt_chunk(extract.features_chunk_df)## label step_major step_minor label_minor
## 2 extract.features_factorize.str.vars 2 0 0
## 1 extract.features_bgn 1 0 0
## bgn end elapsed duration
## 2 27.789 27.850 0.061 0.061
## 1 27.773 27.788 0.015 0.015
## [1] "Total Elapsed Time: 27.85 secs"
# if (glb_save_envir)
# save(glb_feats_df,
# glb_allobs_df, #glb_trnobs_df, glb_fitobs_df, glb_OOBobs_df, glb_newobs_df,
# file=paste0(glb_out_pfx, "extract_features_dsk.RData"))
# load(paste0(glb_out_pfx, "extract_features_dsk.RData"))
replay.petrisim(pn=glb_analytics_pn,
replay.trans=(glb_analytics_avl_objs <- c(glb_analytics_avl_objs,
"data.training.all","data.new")), flip_coord=TRUE)## time trans "bgn " "fit.data.training.all " "predict.data.new " "end "
## 0.0000 multiple enabled transitions: data.training.all data.new model.selected firing: data.training.all
## 1.0000 1 2 1 0 0
## 1.0000 multiple enabled transitions: data.training.all data.new model.selected model.final data.training.all.prediction firing: data.new
## 2.0000 2 1 1 1 0
glb_chunks_df <- myadd_chunk(glb_chunks_df, "cluster.data", major.inc=TRUE)## label step_major step_minor label_minor bgn end elapsed
## 5 extract.features 3 0 0 27.765 29.087 1.322
## 6 cluster.data 4 0 0 29.087 NA NAStep 4.0: cluster data
glb_chunks_df <- myadd_chunk(glb_chunks_df, "manage.missing.data", major.inc=FALSE)## label step_major step_minor label_minor bgn end
## 6 cluster.data 4 0 0 29.087 29.529
## 7 manage.missing.data 4 1 1 29.530 NA
## elapsed
## 6 0.442
## 7 NA# If mice crashes with error: Error in get(as.character(FUN), mode = "function", envir = envir) : object 'State' of mode 'function' was not found
# consider excluding 'State' as a feature
# print(sapply(names(glb_trnobs_df), function(col) sum(is.na(glb_trnobs_df[, col]))))
# print(sapply(names(glb_newobs_df), function(col) sum(is.na(glb_newobs_df[, col]))))
# glb_trnobs_df <- na.omit(glb_trnobs_df)
# glb_newobs_df <- na.omit(glb_newobs_df)
# df[is.na(df)] <- 0
mycheck_problem_data(glb_allobs_df, fctrMaxUniqVals = glbFctrMaxUniqVals)## [1] "numeric data missing in : "
## sold sold.fctr
## 798 798
## [1] "numeric data w/ 0s in : "
## biddable sold cellular.fctr
## 1444 999 1597
## [1] "numeric data w/ Infs in : "
## named integer(0)
## [1] "numeric data w/ NaNs in : "
## named integer(0)
## [1] "string data missing in : "
## description condition cellular carrier color storage
## 1520 0 0 0 0 0
## productline .grpid descr.my
## 0 NA 1520# glb_allobs_df <- na.omit(glb_allobs_df)
# Not refactored into mydsutils.R since glb_*_df might be reassigned
glb_impute_missing_data <- function() {
require(mice)
set.seed(glb_mice_complete.seed)
inp_impent_df <- glb_allobs_df[, setdiff(names(glb_allobs_df),
union(glb_exclude_vars_as_features, glb_rsp_var))]
print("Summary before imputation: ")
print(summary(inp_impent_df))
out_impent_df <- complete(mice(inp_impent_df))
print(summary(out_impent_df))
ret_vars <- sapply(names(out_impent_df),
function(col) ifelse(!identical(out_impent_df[, col],
inp_impent_df[, col]),
col, ""))
ret_vars <- ret_vars[ret_vars != ""]
# complete(mice()) changes attributes of factors even though values don't change
for (col in ret_vars) {
if (inherits(out_impent_df[, col], "factor")) {
if (identical(as.numeric(out_impent_df[, col]),
as.numeric(inp_impent_df[, col])))
ret_vars <- setdiff(ret_vars, col)
}
}
return(out_impent_df[, ret_vars])
}
if (glb_impute_na_data &&
(length(myfind_numerics_missing(glb_allobs_df)) > 0) &&
(ncol(nonna_df <- glb_impute_missing_data()) > 0)) {
for (col in names(nonna_df)) {
glb_allobs_df[, paste0(col, ".nonNA")] <- nonna_df[, col]
glb_exclude_vars_as_features <- c(glb_exclude_vars_as_features, col)
}
}
mycheck_problem_data(glb_allobs_df, fctrMaxUniqVals = glbFctrMaxUniqVals, terminate = TRUE)## [1] "numeric data missing in : "
## sold sold.fctr
## 798 798
## [1] "numeric data w/ 0s in : "
## biddable sold cellular.fctr
## 1444 999 1597
## [1] "numeric data w/ Infs in : "
## named integer(0)
## [1] "numeric data w/ NaNs in : "
## named integer(0)
## [1] "string data missing in : "
## description condition cellular carrier color storage
## 1520 0 0 0 0 0
## productline .grpid descr.my
## 0 NA 1520Step 4.1: manage missing data
mycompute_entropy_df <- function(obs_df, entropy_var, by_var=NULL) {
require(lazyeval)
require(dplyr)
require(tidyr)
if (is.null(by_var)) {
by_var <- ".default"
obs_df$.default <- as.factor(".default")
}
if (!any(grepl(".clusterid", names(obs_df), fixed=TRUE)))
obs_df$.clusterid <- 1
cluster_df <- obs_df %>%
count_(c(by_var, ".clusterid", entropy_var)) %>%
dplyr::filter(n > 0) %>%
dplyr::filter_(interp(~(!is.na(var)), var=as.name(entropy_var))) %>%
unite_(paste0(by_var, ".clusterid"),
c(interp(by_var), ".clusterid")) %>%
spread_(interp(entropy_var), "n", fill=0)
# head(cluster_df)
# sum(cluster_df$n)
tmp.entropy <- sapply(1:nrow(cluster_df),
function(row) entropy(as.numeric(cluster_df[row, -1]), method="ML"))
tmp.knt <- sapply(1:nrow(cluster_df),
function(row) sum(as.numeric(cluster_df[row, -1])))
cluster_df$.entropy <- tmp.entropy; cluster_df$.knt <- tmp.knt
#print(cluster_df)
return(cluster_df)
}
if (glb_cluster) {
require(proxy)
#require(hash)
require(dynamicTreeCut)
require(entropy)
require(tidyr)
mywgtdcosine_dist <- function(x, y=NULL, weights=NULL) {
if (!inherits(x, "matrix"))
x <- as.matrix(x)
if (is.null(weights))
weights <- rep(1, ncol(x))
wgtsx <- matrix(rep(weights / sum(weights), nrow(x)), nrow=nrow(x), byrow=TRUE)
wgtdx <- x * wgtsx
wgtdxsqsum <- as.matrix(rowSums((x ^ 2) * wgtsx), byrow=FALSE)
denom <- sqrt(wgtdxsqsum %*% t(wgtdxsqsum))
ret_mtrx <- 1 - ((sum(weights) ^ 1) * (wgtdx %*% t(wgtdx)) / denom)
ret_mtrx[is.nan(ret_mtrx)] <- 1
diag(ret_mtrx) <- 0
return(ret_mtrx)
}
#pr_DB$delete_entry("mywgtdcosine");
# Need to do this only once across runs ?
if (!pr_DB$entry_exists("mywgtdcosine")) {
pr_DB$set_entry(FUN = mywgtdcosine_dist, names = c("mywgtdcosine"))
pr_DB$modify_entry(names="mywgtdcosine", type="metric", loop=FALSE)
}
#pr_DB$get_entry("mywgtdcosine")
# glb_hash <- hash(key=unique(glb_allobs_df$myCategory),
# values=1:length(unique(glb_allobs_df$myCategory)))
# glb_hash_lst <- hash(key=unique(glb_allobs_df$myCategory),
# values=1:length(unique(glb_allobs_df$myCategory)))
#stop(here"); glb_to_sav(); glb_allobs_df <- sav_allobs_df
cluster_vars <- grep(paste0("[",
toupper(paste0(substr(glb_txt_vars, 1, 1), collapse="")),
"]\\.[PT]\\."),
names(glb_allobs_df), value=TRUE)
# Assign correlations with rsp_var as weights for cosine distance
print("Clustering features: ")
cluster_vars_df <- data.frame(abs.cor.y=abs(cor(
glb_allobs_df[glb_allobs_df$.src == "Train", cluster_vars],
glb_allobs_df[glb_allobs_df$.src == "Train", glb_rsp_var],
use="pairwise.complete.obs")))
print(tail(cluster_vars_df <- orderBy(~ abs.cor.y, subset(cluster_vars_df, !is.na(abs.cor.y))), 5))
print(sprintf(" .rnorm cor: %0.4f",
cor(glb_allobs_df[glb_allobs_df$.src == "Train", ".rnorm"],
glb_allobs_df[glb_allobs_df$.src == "Train", glb_rsp_var], use="pairwise.complete.obs")))
print(sprintf("glb_allobs_df Entropy: %0.4f",
allobs_ent <- entropy(table(glb_allobs_df[, glb_cluster_entropy_var]),
method="ML")))
print(category_df <- mycompute_entropy_df(obs_df=glb_allobs_df,
entropy_var=glb_cluster_entropy_var,
by_var=glb_category_var))
print(sprintf("glb_allobs_df$%s Entropy: %0.4f (%0.4f pct)",
glb_category_var,
category_ent <- weighted.mean(category_df$.entropy, category_df$.knt),
100 * category_ent / allobs_ent))
glb_allobs_df$.clusterid <- 1
#print(max(table(glb_allobs_df$myCategory.fctr) / 20))
grp_ids <- sort(unique(glb_allobs_df[, glb_category_var]))
glb_cluster_size_df_lst <- list()
for (grp in grp_ids) {
# if (grep(grp, levels(grp_ids)) <= 6) next
# if (grep(grp, levels(grp_ids)) > 9) next
# if (grep(grp, levels(grp_ids)) != 10) next
print(sprintf("Category: %s", grp))
ctgry_allobs_df <- glb_allobs_df[glb_allobs_df[, glb_category_var] == grp, ]
if (!inherits(ctgry_allobs_df[, glb_cluster_entropy_var], "factor"))
ctgry_allobs_df[, glb_cluster_entropy_var] <-
as.factor(ctgry_allobs_df[, glb_cluster_entropy_var])
#dstns_dist <- proxy::dist(ctgry_allobs_df[, cluster_vars], method = "cosine")
dstns_dist <- proxy::dist(ctgry_allobs_df[, row.names(cluster_vars_df)],
method = "mywgtdcosine",
weights=cluster_vars_df$abs.cor.y)
# Custom distance functions return a crossdist object
#dstns_mtrx <- as.matrix(dstns_dist)
dstns_mtrx <- matrix(as.vector(dstns_dist), nrow=attr(dstns_dist, "dim")[1],
dimnames=attr(dstns_dist, "dimnames"))
dstns_dist <- as.dist(dstns_mtrx)
print(sprintf("max distance(%0.4f) pair:", max(dstns_mtrx)))
# print(dim(dstns_mtrx))
# print(sprintf("which.max: %d", which.max(dstns_mtrx)))
row_ix <- ceiling(which.max(dstns_mtrx) / ncol(dstns_mtrx))
col_ix <- which.max(dstns_mtrx[row_ix, ])
# print(sprintf("row_ix: %d", row_ix)); print(sprintf("col_ix: %d", col_ix));
# print(dim(ctgry_allobs_df))
print(ctgry_allobs_df[c(row_ix, col_ix),
c(glb_id_var, glb_cluster_entropy_var, glb_category_var, glb_txt_vars, cluster_vars)])
min_dstns_mtrx <- dstns_mtrx
diag(min_dstns_mtrx) <- 1
# Float representations issue -2.22e-16 vs. 0.0000
print(sprintf("min distance(%0.4f) pair:", min(min_dstns_mtrx)))
row_ix <- ceiling(which.min(min_dstns_mtrx) / ncol(min_dstns_mtrx))
col_ix <- which.min(min_dstns_mtrx[row_ix, ])
print(ctgry_allobs_df[c(row_ix, col_ix),
c(glb_id_var, glb_cluster_entropy_var, glb_category_var, glb_txt_vars,
cluster_vars)])
set.seed(glb_cluster.seed)
clusters <- hclust(dstns_dist, method = "ward.D2")
# Workaround to avoid "Error in cutree(dendro, h = heightcutoff) : the 'height' component of 'tree' is not sorted (increasingly)"
if (with(clusters,all.equal(height,sort(height))))
clusters$height <- round(clusters$height,6)
myplclust(clusters, lab=ctgry_allobs_df[, glb_id_var],
lab.col=unclass(ctgry_allobs_df[, glb_cluster_entropy_var]))
opt_minclustersize_df <- data.frame(minclustersize=nrow(ctgry_allobs_df),
entropy=entropy(table(ctgry_allobs_df[, glb_cluster_entropy_var]),
method="ML"))
for (minclustersize in
as.integer(seq(nrow(ctgry_allobs_df) / 2, nrow(ctgry_allobs_df) / 10, length=5))) {
clusterGroups <- cutreeDynamic(clusters, minClusterSize=minclustersize, method="tree",
deepSplit=0)
# Unassigned groups are labeled 0; the largest group has label 1
clusterGroups[clusterGroups == 0] <- 1
ctgry_allobs_df$.clusterid <- clusterGroups
ctgry_clstrs_df <- mycompute_entropy_df(ctgry_allobs_df, glb_cluster_entropy_var)
opt_minclustersize_df <- rbind(opt_minclustersize_df,
data.frame(minclustersize=minclustersize,
entropy=weighted.mean(ctgry_clstrs_df$.entropy, ctgry_clstrs_df$.knt)))
}
opt_minclustersize <-
opt_minclustersize_df$minclustersize[which.min(opt_minclustersize_df$entropy)]
opt_minclustersize_df$.color <-
ifelse(opt_minclustersize_df$minclustersize == opt_minclustersize,
"red", "blue")
print(ggplot(data=opt_minclustersize_df, mapping=aes(x=minclustersize, y=entropy)) +
geom_point(aes(color=.color)) + scale_color_identity() + guides(color = "none") +
geom_line())
glb_cluster_size_df_lst[[grp]] <- opt_minclustersize_df
# select minclustersize that minimizes entropy
clusterGroups <- cutreeDynamic(clusters, minClusterSize=opt_minclustersize,
method="tree",
deepSplit=0)
# Unassigned groups are labeled 0; the largest group has label 1
table(clusterGroups, ctgry_allobs_df[, glb_cluster_entropy_var], useNA="ifany")
clusterGroups[clusterGroups == 0] <- 1
table(clusterGroups, ctgry_allobs_df[, glb_cluster_entropy_var], useNA="ifany")
glb_allobs_df[glb_allobs_df[, glb_category_var]==grp,]$.clusterid <- clusterGroups
}
#all.equal(sav_allobs_df_clusterid, glb_allobs_df$.clusterid)
print(cluster_df <- mycompute_entropy_df(obs_df=glb_allobs_df,
entropy_var=glb_cluster_entropy_var,
by_var=glb_category_var))
print(sprintf("glb_allobs_df$%s$.clusterid Entropy: %0.4f (%0.4f pct)",
glb_category_var,
cluster_ent <- weighted.mean(cluster_df$.entropy, cluster_df$.knt),
100 * cluster_ent / category_ent))
glb_allobs_df$.clusterid.fctr <- as.factor(glb_allobs_df$.clusterid)
# .clusterid.fctr is created automatically (probably ?) later
glb_exclude_vars_as_features <- c(glb_exclude_vars_as_features, ".clusterid")
if (!is.null(glb_category_var))
# glb_interaction_only_feats_lst[ifelse(grepl("\\.fctr", glb_category_var),
# glb_category_var,
# paste0(glb_category_var, ".fctr"))] <-
# c(".clusterid.fctr")
glb_interaction_only_feats_lst[[".clusterid.fctr"]] <-
ifelse(grepl("\\.fctr", glb_category_var), glb_category_var,
paste0(glb_category_var, ".fctr"))
if (glb_exclude_cluster_vars_as_features)
glb_exclude_vars_as_features <- c(glb_exclude_vars_as_features,
cluster_vars)
}
# Last call for data modifications
#stop(here") # sav_allobs_df <- glb_allobs_df
# glb_allobs_df[(glb_allobs_df$PropR == 0.75) & (glb_allobs_df$State == "Hawaii"), "PropR.fctr"] <- "N"
# Re-partition
glb_trnobs_df <- subset(glb_allobs_df, .src == "Train")
glb_newobs_df <- subset(glb_allobs_df, .src == "Test")
glb_chunks_df <- myadd_chunk(glb_chunks_df, "partition.data.training", major.inc=TRUE)## label step_major step_minor label_minor bgn end
## 7 manage.missing.data 4 1 1 29.530 29.623
## 8 partition.data.training 5 0 0 29.624 NA
## elapsed
## 7 0.093
## 8 NAStep 5.0: partition data training
if (all(is.na(glb_newobs_df[, glb_rsp_var]))) {
set.seed(glb_split_sample.seed)
OOB_size <- nrow(glb_newobs_df) * 1.1
if (is.null(glb_category_var)) {
require(caTools)
split <- sample.split(glb_trnobs_df[, glb_rsp_var_raw],
SplitRatio=OOB_size / nrow(glb_trnobs_df))
glb_OOBobs_df <- glb_trnobs_df[split ,]
glb_fitobs_df <- glb_trnobs_df[!split, ]
} else {
sample_vars <- c(glb_rsp_var_raw, glb_category_var)
rspvar_freq_df <- orderBy(reformulate(glb_rsp_var_raw),
mycreate_sqlxtab_df(glb_trnobs_df, glb_rsp_var_raw))
OOB_rspvar_size <- 1.0 * OOB_size * rspvar_freq_df$.n / sum(rspvar_freq_df$.n)
newobs_freq_df <- orderBy(reformulate(glb_category_var),
mycreate_sqlxtab_df(glb_newobs_df, glb_category_var))
trnobs_freq_df <- orderBy(reformulate(glb_category_var),
mycreate_sqlxtab_df(glb_trnobs_df, glb_category_var))
allobs_freq_df <- merge(newobs_freq_df, trnobs_freq_df, by=glb_category_var,
all=TRUE, sort=TRUE, suffixes=c(".Tst", ".Train"))
allobs_freq_df[is.na(allobs_freq_df)] <- 0
OOB_strata_size <- ceiling(
as.vector(matrix(allobs_freq_df$.n.Tst * 1.0 / sum(allobs_freq_df$.n.Tst)) %*%
matrix(OOB_rspvar_size, nrow = 1)))
OOB_strata_size[OOB_strata_size == 0] <- 1
OOB_strata_df <- expand.grid(glb_rsp_var_raw=rspvar_freq_df[, glb_rsp_var_raw],
glb_category_var=allobs_freq_df[, glb_category_var])
names(OOB_strata_df) <- sample_vars
OOB_strata_df <- orderBy(reformulate(sample_vars), OOB_strata_df)
trnobs_univ_df <- orderBy(reformulate(sample_vars),
mycreate_sqlxtab_df(glb_trnobs_df, sample_vars))
trnobs_univ_df <- merge(trnobs_univ_df, OOB_strata_df, all=TRUE)
tmp_trnobs_df <- orderBy(reformulate(c(glb_rsp_var_raw, glb_category_var)),
glb_trnobs_df)
# Adjust OOB_strata_size (desired # of OOB obs) if > # of trn obs
ix <- which(!is.na(trnobs_univ_df$.n) & (OOB_strata_size > trnobs_univ_df$.n))
OOB_strata_size[ix] <- trnobs_univ_df[ix, ".n"]
require(sampling)
split_strata <- sampling::strata(tmp_trnobs_df,
stratanames = c(glb_rsp_var_raw, glb_category_var),
size = OOB_strata_size[!is.na(trnobs_univ_df$.n)],
method = "srswor")
glb_OOBobs_df <- getdata(tmp_trnobs_df, split_strata)[, names(glb_trnobs_df)]
glb_fitobs_df <- glb_trnobs_df[!glb_trnobs_df[, glb_id_var] %in%
glb_OOBobs_df[, glb_id_var], ]
}
} else {
print(sprintf("Newdata contains non-NA data for %s; setting OOB to Newdata",
glb_rsp_var))
glb_fitobs_df <- glb_trnobs_df; glb_OOBobs_df <- glb_newobs_df
}## Loading required package: sampling
##
## Attaching package: 'sampling'
##
## The following objects are masked from 'package:survival':
##
## cluster, strata
##
## The following object is masked from 'package:caret':
##
## clusterif (!is.null(glb_max_fitobs) && (nrow(glb_fitobs_df) > glb_max_fitobs)) {
warning("glb_fitobs_df restricted to glb_max_fitobs: ",
format(glb_max_fitobs, big.mark=","))
org_fitobs_df <- glb_fitobs_df
glb_fitobs_df <-
org_fitobs_df[split <- sample.split(org_fitobs_df[, glb_rsp_var_raw],
SplitRatio=glb_max_fitobs), ]
org_fitobs_df <- NULL
}
if (!is.null(glb_obsfit_outliers)) {
glb_OOBobs_df <- rbind(glb_OOBobs_df,
glb_fitobs_df[glb_fitobs_df[, glb_id_var] %in% glb_obsfit_outliers, ])
glb_fitobs_df <- glb_fitobs_df[!(glb_fitobs_df[, glb_id_var] %in% glb_obsfit_outliers), ]
}
glb_allobs_df$.lcn <- ""; glb_trnobs_df$.lcn <- "";
glb_allobs_df[glb_allobs_df[, glb_id_var] %in%
glb_fitobs_df[, glb_id_var], ".lcn"] <- "Fit"
glb_trnobs_df[glb_trnobs_df[, glb_id_var] %in%
glb_fitobs_df[, glb_id_var], ".lcn"] <- "Fit"
glb_allobs_df[glb_allobs_df[, glb_id_var] %in%
glb_OOBobs_df[, glb_id_var], ".lcn"] <- "OOB"
glb_trnobs_df[glb_trnobs_df[, glb_id_var] %in%
glb_OOBobs_df[, glb_id_var], ".lcn"] <- "OOB"
dsp_class_dstrb <- function(obs_df, location_var, partition_var) {
xtab_df <- mycreate_xtab_df(obs_df, c(location_var, partition_var))
rownames(xtab_df) <- xtab_df[, location_var]
xtab_df <- xtab_df[, -grepl(location_var, names(xtab_df))]
print(xtab_df)
print(xtab_df / rowSums(xtab_df, na.rm=TRUE))
}
# Ensure proper splits by glb_rsp_var_raw & user-specified feature for OOB vs. new
if (!is.null(glb_category_var)) {
if (glb_is_classification)
dsp_class_dstrb(glb_allobs_df, ".lcn", glb_rsp_var_raw)
newobs_ctgry_df <- mycreate_sqlxtab_df(subset(glb_allobs_df, .src == "Test"),
glb_category_var)
OOBobs_ctgry_df <- mycreate_sqlxtab_df(subset(glb_allobs_df, .lcn == "OOB"),
glb_category_var)
glb_ctgry_df <- merge(newobs_ctgry_df, OOBobs_ctgry_df, by=glb_category_var
, all=TRUE, suffixes=c(".Tst", ".OOB"))
glb_ctgry_df$.freqRatio.Tst <- glb_ctgry_df$.n.Tst / sum(glb_ctgry_df$.n.Tst, na.rm=TRUE)
glb_ctgry_df$.freqRatio.OOB <- glb_ctgry_df$.n.OOB / sum(glb_ctgry_df$.n.OOB, na.rm=TRUE)
print(orderBy(~-.freqRatio.Tst-.freqRatio.OOB, glb_ctgry_df))
}## sold.0 sold.1 sold.NA
## NA NA 798
## Fit 515 444 NA
## OOB 484 416 NA
## sold.0 sold.1 sold.NA
## NA NA 1
## Fit 0.5370177 0.4629823 NA
## OOB 0.5377778 0.4622222 NA
## prdl.descr.my.fctr .n.Tst .n.OOB .freqRatio.Tst .freqRatio.OOB
## 5 iPad2#0 83 93 0.10401003 0.103333333
## 6 iPad2#1 71 79 0.08897243 0.087777778
## 16 iPadmini#0 65 73 0.08145363 0.081111111
## 2 Unknown#1 47 52 0.05889724 0.057777778
## 14 iPadAir2#0 47 52 0.05889724 0.057777778
## 3 iPad1#0 46 52 0.05764411 0.057777778
## 17 iPadmini#1 46 52 0.05764411 0.057777778
## 1 Unknown#0 45 50 0.05639098 0.055555556
## 4 iPad1#1 42 47 0.05263158 0.052222222
## 12 iPadAir#0 41 46 0.05137845 0.051111111
## 10 iPad4#1 39 44 0.04887218 0.048888889
## 18 iPadmini2#0 35 39 0.04385965 0.043333333
## 13 iPadAir#1 33 37 0.04135338 0.041111111
## 7 iPad3#0 30 34 0.03759398 0.037777778
## 9 iPad4#0 29 33 0.03634085 0.036666667
## 20 iPadmini3#0 29 33 0.03634085 0.036666667
## 8 iPad3#1 25 28 0.03132832 0.031111111
## 19 iPadmini2#1 21 24 0.02631579 0.026666667
## 15 iPadAir2#1 15 17 0.01879699 0.018888889
## 21 iPadmini3#1 9 10 0.01127820 0.011111111
## 22 iPadminiRetina#0 NA 2 NA 0.002222222
## 23 iPadminiRetina#1 NA 2 NA 0.002222222
## 11 iPad5#0 NA 1 NA 0.001111111print("glb_allobs_df: "); print(dim(glb_allobs_df))## [1] "glb_allobs_df: "## [1] 2657 23print("glb_trnobs_df: "); print(dim(glb_trnobs_df))## [1] "glb_trnobs_df: "## [1] 1859 23print("glb_fitobs_df: "); print(dim(glb_fitobs_df))## [1] "glb_fitobs_df: "## [1] 959 22print("glb_OOBobs_df: "); print(dim(glb_OOBobs_df))## [1] "glb_OOBobs_df: "## [1] 900 22print("glb_newobs_df: "); print(dim(glb_newobs_df))## [1] "glb_newobs_df: "## [1] 798 22# # Does not handle NULL or length(glb_id_var) > 1
if (glb_save_envir)
save(glb_allobs_df, #glb_trnobs_df, glb_fitobs_df, glb_OOBobs_df, glb_newobs_df,
file=paste0(glb_out_pfx, "blddfs_dsk.RData"))
# load(paste0(glb_out_pfx, "blddfs_dsk.RData"))
rm(split)## Warning in rm(split): object 'split' not foundglb_chunks_df <- myadd_chunk(glb_chunks_df, "select.features", major.inc=TRUE)## label step_major step_minor label_minor bgn end
## 8 partition.data.training 5 0 0 29.624 30.402
## 9 select.features 6 0 0 30.402 NA
## elapsed
## 8 0.778
## 9 NAStep 6.0: select features
#stop(here"); glb_to_sav(); glb_allobs_df <- sav_allobs_df
print(glb_feats_df <- myselect_features(entity_df=glb_trnobs_df,
exclude_vars_as_features=glb_exclude_vars_as_features,
rsp_var=glb_rsp_var))## id cor.y exclude.as.feat
## sold sold 1.000000000 1
## biddable biddable 0.548178838 0
## startprice startprice -0.456976721 0
## UniqueID UniqueID -0.189546626 1
## condition.fctr condition.fctr -0.153549007 0
## cellular.fctr cellular.fctr -0.073944342 0
## carrier.fctr carrier.fctr -0.059689081 0
## prdl.descr.my.fctr prdl.descr.my.fctr -0.057328786 0
## color.fctr color.fctr -0.040738354 0
## storage.fctr storage.fctr -0.011948513 0
## .rnorm .rnorm -0.001435011 0
## cor.y.abs
## sold 1.000000000
## biddable 0.548178838
## startprice 0.456976721
## UniqueID 0.189546626
## condition.fctr 0.153549007
## cellular.fctr 0.073944342
## carrier.fctr 0.059689081
## prdl.descr.my.fctr 0.057328786
## color.fctr 0.040738354
## storage.fctr 0.011948513
## .rnorm 0.001435011print(glb_feats_df <- orderBy(~-cor.y,
myfind_cor_features(feats_df=glb_feats_df, obs_df=glb_trnobs_df, rsp_var=glb_rsp_var,
nzv.freqCut=glb_nzv_freqCut, nzv.uniqueCut=glb_nzv_uniqueCut)))## [1] "cor(carrier.fctr, cellular.fctr)=0.7138"
## [1] "cor(sold.fctr, carrier.fctr)=-0.0597"
## [1] "cor(sold.fctr, cellular.fctr)=-0.0739"## Warning in myfind_cor_features(feats_df = glb_feats_df, obs_df =
## glb_trnobs_df, : Identified carrier.fctr as highly correlated with
## cellular.fctr## id cor.y exclude.as.feat
## sold sold 1.000000000 1
## biddable biddable 0.548178838 0
## .rnorm .rnorm -0.001435011 0
## storage.fctr storage.fctr -0.011948513 0
## color.fctr color.fctr -0.040738354 0
## prdl.descr.my.fctr prdl.descr.my.fctr -0.057328786 0
## carrier.fctr carrier.fctr -0.059689081 0
## cellular.fctr cellular.fctr -0.073944342 0
## condition.fctr condition.fctr -0.153549007 0
## UniqueID UniqueID -0.189546626 1
## startprice startprice -0.456976721 0
## cor.y.abs cor.high.X freqRatio percentUnique
## sold 1.000000000 <NA> 1.161628 0.1075847
## biddable 0.548178838 <NA> 1.221027 0.1075847
## .rnorm 0.001435011 <NA> 1.000000 100.0000000
## storage.fctr 0.011948513 <NA> 2.741176 0.2689618
## color.fctr 0.040738354 <NA> 1.574610 0.2689618
## prdl.descr.my.fctr 0.057328786 <NA> 1.161074 1.2372243
## carrier.fctr 0.059689081 cellular.fctr 3.195965 0.3765465
## cellular.fctr 0.073944342 <NA> 2.112381 0.1613771
## condition.fctr 0.153549007 <NA> 4.003460 0.3227542
## UniqueID 0.189546626 <NA> 1.000000 100.0000000
## startprice 0.456976721 <NA> 2.807692 30.1775148
## zeroVar nzv is.cor.y.abs.low
## sold FALSE FALSE FALSE
## biddable FALSE FALSE FALSE
## .rnorm FALSE FALSE FALSE
## storage.fctr FALSE FALSE FALSE
## color.fctr FALSE FALSE FALSE
## prdl.descr.my.fctr FALSE FALSE FALSE
## carrier.fctr FALSE FALSE FALSE
## cellular.fctr FALSE FALSE FALSE
## condition.fctr FALSE FALSE FALSE
## UniqueID FALSE FALSE FALSE
## startprice FALSE FALSE FALSEplt_feats_df <- glb_feats_df
print(myplot_scatter(plt_feats_df, "percentUnique", "freqRatio",
colorcol_name="nzv", jitter=TRUE) +
#geom_point(aes(shape=nzv)) +
geom_point() +
xlim(-5, 25) +
geom_hline(yintercept=glb_nzv_freqCut) +
geom_vline(xintercept=glb_nzv_uniqueCut))## Warning in myplot_scatter(plt_feats_df, "percentUnique", "freqRatio",
## colorcol_name = "nzv", : converting nzv to class:factor## Warning: Removed 3 rows containing missing values (geom_point).## Warning: Removed 3 rows containing missing values (geom_point).## Warning: Removed 3 rows containing missing values (geom_point).
print(subset(glb_feats_df, nzv))## [1] id cor.y exclude.as.feat cor.y.abs
## [5] cor.high.X freqRatio percentUnique zeroVar
## [9] nzv is.cor.y.abs.low
## <0 rows> (or 0-length row.names)tmp_allobs_df <-
glb_allobs_df[, union(setdiff(names(glb_allobs_df), subset(glb_feats_df, nzv)$id),
glb_cluster_entropy_var)]
glb_trnobs_df <- subset(tmp_allobs_df, .src == "Train")
glb_newobs_df <- subset(tmp_allobs_df, .src == "Test")
glb_feats_df$interaction.feat <- NA
for (feat in names(glb_interaction_only_feats_lst))
glb_feats_df[glb_feats_df$id %in% feat, "interaction.feat"] <-
glb_interaction_only_feats_lst[[feat]]
#stop(here"); glb_to_sav(); glb_allobs_df <- sav_allobs_df
indep_vars <- subset(glb_feats_df, !nzv & (exclude.as.feat != 1))[, "id"]
numeric_indep_vars <- indep_vars[!grepl(".fctr", indep_vars, fixed=TRUE)]
glb_feats_df$shapiro.test.p.value <- NA
glb_feats_df[glb_feats_df$id %in% numeric_indep_vars, "shapiro.test.p.value"] <-
sapply(numeric_indep_vars, function(var) shapiro.test(glb_trnobs_df[, var])$p.value)
not_nrml_feats_df <- glb_feats_df %>%
subset(!is.na(shapiro.test.p.value)) %>%
subset((shapiro.test.p.value < 0.05) || (id == ".rnorm")) %>%
arrange(shapiro.test.p.value)
row.names(not_nrml_feats_df) <- not_nrml_feats_df$id
#plt_trnobs_df <- glb_trnobs_df[, c("D.npnct05.log", ".rnorm")]
plt_trnobs_df <- glb_trnobs_df[, c(union(not_nrml_feats_df$id[1:min(5, nrow(not_nrml_feats_df))],
".rnorm"), glb_cluster_entropy_var)]
print(myplot_violin(plt_trnobs_df, setdiff(names(plt_trnobs_df), glb_cluster_entropy_var),
xcol_name = glb_cluster_entropy_var) +
facet_wrap(~variable, scales="free"))
#myplot_histogram(plt_trnobs_df, "D.npnct11.log", fill_col_name="sold", show_stats = TRUE)
myadjust_interaction_feats <- function(vars_vctr) {
for (feat in subset(glb_feats_df, !is.na(interaction.feat))$id)
if (feat %in% vars_vctr)
vars_vctr <- union(setdiff(vars_vctr, feat),
paste0(glb_feats_df[glb_feats_df$id == feat, "interaction.feat"], ":",
feat))
return(vars_vctr)
}
# shd .clusterid.fctr be excluded from this ? or include encoding of glb_category_var:.clusterid.fctr ?
indep_vars <-
myadjust_interaction_feats(subset(glb_feats_df, !nzv & (exclude.as.feat != 1))[, "id"])
myrun_rfe <- function(obs_df, indep_vars, sizes=NULL) {
rfe_obs_df <- myget_vectorized_obs_df(obs_df, glb_rsp_var, indep_vars)
predictors_vctr <- setdiff(names(rfe_obs_df), glb_rsp_var)
if (is.null(sizes))
sizes <- tail(2^(1:as.integer(log2(length(predictors_vctr)))), 5)
rfe_control <- rfeControl(functions=rfFuncs, method="repeatedcv", number=glb_rcv_n_folds,
repeats=glb_rcv_n_repeats, verbose=TRUE, returnResamp = "all",
seeds=mygen_seeds(seeds_lst_len=(glb_rcv_n_folds * glb_rcv_n_repeats) + 1,
seeds_elmnt_lst_len=(length(sizes) + 1)))
set.seed(113)
rfe_results <- rfe(rfe_obs_df[, predictors_vctr],
rfe_obs_df[, glb_rsp_var],
sizes=sizes, metric=unlist(strsplit(glb_model_evl_criteria, "[.]"))[2],
maximize=ifelse(unlist(strsplit(glb_model_evl_criteria, "[.]"))[1] == "max",
TRUE, FALSE),
rfeControl=rfe_control)
print(rfe_results)
print(predictors(rfe_results))
# print(plot(rfe_results, type=c("g", "o")))
# print(plot(rfe_results))
print(ggplot(rfe_results))
return(rfe_results)
}
rfe_fit_results <- myrun_rfe(glb_fitobs_df, indep_vars, glb_rfe_fit_sizes)## Warning in rfe.default(rfe_obs_df[, predictors_vctr], rfe_obs_df[,
## glb_rsp_var], : Metric 'auc' is not created by the summary function;
## 'Accuracy' will be used instead##
## Recursive feature selection
##
## Outer resampling method: Cross-Validated (3 fold, repeated 3 times)
##
## Resampling performance over subset size:
##
## Variables Accuracy Kappa AccuracySD KappaSD Selected
## 2 0.7842 0.5620 0.02634 0.05287
## 4 0.7835 0.5604 0.01896 0.03886
## 8 0.7873 0.5684 0.02452 0.04945
## 16 0.7908 0.5751 0.02504 0.05086
## 32 0.7977 0.5903 0.02178 0.04394
## 59 0.8040 0.6028 0.02053 0.04151 *
##
## The top 5 variables (out of 59):
## startprice, biddable, prdl.descr.my.fctriPad4#1, prdl.descr.my.fctriPad2#0, condition.fctrNew
##
## [1] "startprice"
## [2] "biddable"
## [3] "prdl.descr.my.fctriPad4#1"
## [4] "prdl.descr.my.fctriPad2#0"
## [5] "condition.fctrNew"
## [6] "prdl.descr.my.fctriPad3#0"
## [7] "condition.fctrFor parts or not working"
## [8] "cellular.fctrUnknown"
## [9] "prdl.descr.my.fctriPadmini3#0"
## [10] "cellular.fctrUnknown:carrier.fctrUnknown"
## [11] "storage.fctr128"
## [12] "storage.fctr16"
## [13] "storage.fctrUnknown"
## [14] "prdl.descr.my.fctriPad1#0"
## [15] "prdl.descr.my.fctriPad2#1"
## [16] "storage.fctr32"
## [17] "prdl.descr.my.fctriPadAir#1"
## [18] "prdl.descr.my.fctriPad4#0"
## [19] "prdl.descr.my.fctriPadAir#0"
## [20] "cellular.fctr0:carrier.fctrNone"
## [21] "cellular.fctr1"
## [22] "prdl.descr.my.fctrUnknown#1"
## [23] "condition.fctrSeller refurbished"
## [24] "cellular.fctr1:carrier.fctrVerizon"
## [25] "prdl.descr.my.fctriPadmini2#0"
## [26] "prdl.descr.my.fctriPadAir2#0"
## [27] "prdl.descr.my.fctriPadminiRetina#0"
## [28] "prdl.descr.my.fctriPadmini#0"
## [29] ".rnorm"
## [30] "storage.fctr64"
## [31] "prdl.descr.my.fctriPad1#1"
## [32] "condition.fctrNew other (see details)"
## [33] "prdl.descr.my.fctriPadmini#1"
## [34] "cellular.fctr1:carrier.fctrT-Mobile"
## [35] "color.fctrWhite"
## [36] "prdl.descr.my.fctriPadmini3#1"
## [37] "prdl.descr.my.fctriPadAir2#1"
## [38] "color.fctrUnknown"
## [39] "cellular.fctr1:carrier.fctrUnknown"
## [40] "color.fctrSpace Gray"
## [41] "cellular.fctr0:carrier.fctrOther"
## [42] "cellular.fctr0:carrier.fctrSprint"
## [43] "cellular.fctr0:carrier.fctrT-Mobile"
## [44] "cellular.fctr0:carrier.fctrUnknown"
## [45] "cellular.fctr0:carrier.fctrVerizon"
## [46] "cellular.fctr1:carrier.fctrNone"
## [47] "cellular.fctr1:carrier.fctrOther"
## [48] "cellular.fctrUnknown:carrier.fctrNone"
## [49] "cellular.fctrUnknown:carrier.fctrOther"
## [50] "cellular.fctrUnknown:carrier.fctrSprint"
## [51] "cellular.fctrUnknown:carrier.fctrT-Mobile"
## [52] "cellular.fctrUnknown:carrier.fctrVerizon"
## [53] "prdl.descr.my.fctriPad5#0"
## [54] "prdl.descr.my.fctriPadminiRetina#1"
## [55] "condition.fctrManufacturer refurbished"
## [56] "prdl.descr.my.fctriPadmini2#1"
## [57] "color.fctrGold"
## [58] "cellular.fctr1:carrier.fctrSprint"
## [59] "prdl.descr.my.fctriPad3#1"
# print(all.equal(rfe_results[-which(names(rfe_results) == "times")],
# sav_rfe_results[-which(names(sav_rfe_results) == "times")]))
# require(mRMRe)
# indep_vars_vctr <- subset(glb_feats_df, !nzv &
# (exclude.as.feat != 1))[, "id"]
# indep_vars_vctr <- setdiff(indep_vars_vctr,
# myfind_fctr_cols_df(glb_trnobs_df[, c(glb_rsp_var, indep_vars_vctr)]))
# tmp_trnobs_df <- glb_trnobs_df[, c(glb_rsp_var, indep_vars_vctr)]
# tmp_trnobs_df$biddable <- as.numeric(tmp_trnobs_df$biddable)
# dd <- mRMR.data(data = tmp_trnobs_df)
# mRMRe.fltr <- mRMR.classic(data = dd, target_indices = c(1), feature_count = 10)
# print(solutions(mRMRe.fltr)[[1]])
# print(apply(solutions(mRMRe.fltr)[[1]], 2, function(x, y) { return(y[x]) },
# y=featureNames(dd)))
# print(featureNames(dd)[solutions(mRMRe.fltr)[[1]]])
# print(mRMRe.fltr@filters); print(mRMRe.fltr@scores)
mycheck_problem_data(glb_allobs_df, fctrMaxUniqVals = glbFctrMaxUniqVals, terminate = TRUE)## [1] "numeric data missing in : "
## sold sold.fctr
## 798 798
## [1] "numeric data w/ 0s in : "
## biddable sold cellular.fctr
## 1444 999 1597
## [1] "numeric data w/ Infs in : "
## named integer(0)
## [1] "numeric data w/ NaNs in : "
## named integer(0)
## [1] "string data missing in : "
## description condition cellular carrier color storage
## 1520 0 0 0 0 0
## productline .grpid descr.my .lcn
## 0 NA 1520 798# glb_allobs_df %>% filter(is.na(Married.fctr)) %>% tbl_df()
# glb_allobs_df %>% count(Married.fctr)
# levels(glb_allobs_df$Married.fctr)
print("glb_feats_df:"); print(dim(glb_feats_df))## [1] "glb_feats_df:"## [1] 11 12sav_feats_df <- glb_feats_df
glb_feats_df <- sav_feats_df
glb_feats_df[, "rsp_var_raw"] <- FALSE
glb_feats_df[glb_feats_df$id == glb_rsp_var_raw, "rsp_var_raw"] <- TRUE
glb_feats_df$exclude.as.feat <- (glb_feats_df$exclude.as.feat == 1)
if (!is.null(glb_id_var) && glb_id_var != ".rownames")
glb_feats_df[glb_feats_df$id %in% glb_id_var, "id_var"] <- TRUE
add_feats_df <- data.frame(id=glb_rsp_var, exclude.as.feat=TRUE, rsp_var=TRUE)
row.names(add_feats_df) <- add_feats_df$id; print(add_feats_df)## id exclude.as.feat rsp_var
## sold.fctr sold.fctr TRUE TRUEglb_feats_df <- myrbind_df(glb_feats_df, add_feats_df)
if (glb_id_var != ".rownames")
print(subset(glb_feats_df, rsp_var_raw | rsp_var | id_var)) else
print(subset(glb_feats_df, rsp_var_raw | rsp_var)) ## id cor.y exclude.as.feat cor.y.abs cor.high.X
## sold sold 1.0000000 TRUE 1.0000000 <NA>
## UniqueID UniqueID -0.1895466 TRUE 0.1895466 <NA>
## sold.fctr sold.fctr NA TRUE NA <NA>
## freqRatio percentUnique zeroVar nzv is.cor.y.abs.low
## sold 1.161628 0.1075847 FALSE FALSE FALSE
## UniqueID 1.000000 100.0000000 FALSE FALSE FALSE
## sold.fctr NA NA NA NA NA
## interaction.feat shapiro.test.p.value rsp_var_raw id_var rsp_var
## sold <NA> NA TRUE NA NA
## UniqueID <NA> NA FALSE TRUE NA
## sold.fctr <NA> NA NA NA TRUEprint("glb_feats_df vs. glb_allobs_df: "); ## [1] "glb_feats_df vs. glb_allobs_df: "print(setdiff(glb_feats_df$id, names(glb_allobs_df)))## character(0)print("glb_allobs_df vs. glb_feats_df: "); ## [1] "glb_allobs_df vs. glb_feats_df: "# Ensure these are only chr vars
print(setdiff(setdiff(names(glb_allobs_df), glb_feats_df$id),
myfind_chr_cols_df(glb_allobs_df)))## character(0)if (glb_save_envir)
save(glb_feats_df,
glb_allobs_df, #glb_trnobs_df, glb_fitobs_df, glb_OOBobs_df, glb_newobs_df,
file=paste0(glb_out_pfx, "selfts_dsk.RData"))
# load(paste0(glb_out_pfx, "blddfs_dsk.RData"))
# if (!all.equal(tmp_feats_df, glb_feats_df))
# stop("glb_feats_df r/w not working")
# if (!all.equal(tmp_entity_df, glb_allobs_df))
# stop("glb_allobs_df r/w not working")
glb_chunks_df <- myadd_chunk(glb_chunks_df, "fit.models", major.inc=TRUE)## label step_major step_minor label_minor bgn end elapsed
## 9 select.features 6 0 0 30.402 52.896 22.494
## 10 fit.models 7 0 0 52.896 NA NAStep 7.0: fit models
# load(paste0(glb_out_pfx, "dsk.RData"))
get_model_sel_frmla <- function() {
model_evl_terms <- c(NULL)
# min.aic.fit might not be avl
lclMdlEvlCriteria <-
glb_model_evl_criteria[glb_model_evl_criteria %in% names(glb_models_df)]
for (metric in lclMdlEvlCriteria)
model_evl_terms <- c(model_evl_terms,
ifelse(length(grep("max", metric)) > 0, "-", "+"), metric)
if (glb_is_classification && glb_is_binomial)
model_evl_terms <- c(model_evl_terms, "-", "opt.prob.threshold.OOB")
model_sel_frmla <- as.formula(paste(c("~ ", model_evl_terms), collapse = " "))
return(model_sel_frmla)
}
get_dsp_models_df <- function() {
dsp_models_df <-
#orderBy(get_model_sel_frmla(), glb_models_df)[, c("id", glb_model_evl_criteria)]
orderBy(get_model_sel_frmla(), glb_models_df)[, dsp_models_cols]
nCvMdl <- sapply(glb_models_lst, function(mdl) nrow(mdl$results))
nParams <- sapply(glb_models_lst, function(mdl) ifelse(mdl$method == "custom", 0,
nrow(subset(modelLookup(mdl$method), parameter != "parameter"))))
# nCvMdl <- nCvMdl[names(nCvMdl) != "avNNet"]
# nParams <- nParams[names(nParams) != "avNNet"]
if (length(cvMdlProblems <- nCvMdl[nCvMdl <= nParams]) > 0) {
print("Cross Validation issues:")
warning("Cross Validation issues:")
print(cvMdlProblems)
}
pltMdls <- setdiff(names(nCvMdl), names(cvMdlProblems))
pltMdls <- setdiff(pltMdls, names(nParams[nParams == 0]))
# length(pltMdls) == 21
png(paste0(glb_out_pfx, "bestTune.png"), width = 480 * 2, height = 480 * 4)
grid.newpage()
pushViewport(viewport(layout = grid.layout(ceiling(length(pltMdls) / 2.0), 2)))
pltIx <- 1
for (mdlId in pltMdls) {
print(ggplot(glb_models_lst[[mdlId]], highBestTune = TRUE) + labs(title = mdlId),
vp = viewport(layout.pos.row = ceiling(pltIx / 2.0),
layout.pos.col = ((pltIx - 1) %% 2) + 1))
pltIx <- pltIx + 1
}
dev.off()
return(dsp_models_df)
}
#get_dsp_models_df()
if (glb_is_classification && glb_is_binomial &&
(length(unique(glb_fitobs_df[, glb_rsp_var])) < 2))
stop("glb_fitobs_df$", glb_rsp_var, ": contains less than 2 unique values: ",
paste0(unique(glb_fitobs_df[, glb_rsp_var]), collapse=", "))
max_cor_y_x_vars <- orderBy(~ -cor.y.abs,
subset(glb_feats_df, (exclude.as.feat == 0) & !nzv & !is.cor.y.abs.low &
is.na(cor.high.X)))[1:2, "id"]
# while(length(max_cor_y_x_vars) < 2) {
# max_cor_y_x_vars <- c(max_cor_y_x_vars, orderBy(~ -cor.y.abs,
# subset(glb_feats_df, (exclude.as.feat == 0) & !is.cor.y.abs.low))[3, "id"])
# }
#stop(here"); glb_to_sav(); glb_models_lst <- sav_models_lst; glb_models_df <- sav_models_df
if (!is.null(glb_Baseline_mdl_var)) {
if ((max_cor_y_x_vars[1] != glb_Baseline_mdl_var) &
(glb_feats_df[glb_feats_df$id == max_cor_y_x_vars[1], "cor.y.abs"] >
glb_feats_df[glb_feats_df$id == glb_Baseline_mdl_var, "cor.y.abs"]))
stop(max_cor_y_x_vars[1], " has a higher correlation with ", glb_rsp_var,
" than the Baseline var: ", glb_Baseline_mdl_var)
}
glb_model_type <- ifelse(glb_is_regression, "regression", "classification")
# Model specs
c("id.prefix", "method", "type",
# trainControl params
"preProc.method", "cv.n.folds", "cv.n.repeats", "summary.fn",
# train params
"metric", "metric.maximize", "tune.df")## [1] "id.prefix" "method" "type"
## [4] "preProc.method" "cv.n.folds" "cv.n.repeats"
## [7] "summary.fn" "metric" "metric.maximize"
## [10] "tune.df"# Baseline
if (!is.null(glb_Baseline_mdl_var))
ret_lst <- myfit_mdl(mdl_id="Baseline",
model_method="mybaseln_classfr",
indep_vars_vctr=glb_Baseline_mdl_var,
rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)
# Most Frequent Outcome "MFO" model: mean(y) for regression
# Not using caret's nullModel since model stats not avl
# Cannot use rpart for multinomial classification since it predicts non-MFO
ret_lst <- myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst = list(
id.prefix = "MFO", type = glb_model_type, trainControl.method = "none",
train.method = ifelse(glb_is_regression, "lm", "myMFO_classfr"))),
indep_vars = ".rnorm", rsp_var = glb_rsp_var,
fit_df = glb_fitobs_df, OOB_df = glb_OOBobs_df)## [1] "fitting model: MFO.myMFO_classfr"
## [1] " indep_vars: .rnorm"## Warning in if (mdl_specs_lst[["train.method"]] %in% c("bayesglm", "glm"))
## allowPar <- FALSE: the condition has length > 1 and only the first element
## will be used## Fitting parameter = none on full training set
## [1] "in MFO.Classifier$fit"
## [1] "unique.vals:"
## [1] N Y
## Levels: N Y
## [1] "unique.prob:"
## y
## N Y
## 0.5370177 0.4629823
## [1] "MFO.val:"
## [1] "N"
## Length Class Mode
## unique.vals 2 factor numeric
## unique.prob 2 -none- numeric
## MFO.val 1 -none- character
## x.names 1 -none- character
## xNames 1 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 2 -none- character
## [1] " calling mypredict_mdl for fit:"## Loading required package: ROCR
## Loading required package: gplots
##
## Attaching package: 'gplots'
##
## The following object is masked from 'package:stats':
##
## lowess## [1] "in MFO.Classifier$prob"
## N Y
## 1 0.5370177 0.4629823
## 2 0.5370177 0.4629823
## 3 0.5370177 0.4629823
## 4 0.5370177 0.4629823
## 5 0.5370177 0.4629823
## 6 0.5370177 0.4629823
## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.MFO.myMFO_classfr.N
## 1 N 515
## 2 Y 444
## Prediction
## Reference N Y
## N 515 0
## Y 444 0
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 5.370177e-01 0.000000e+00 5.048652e-01 5.689423e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 5.132330e-01 3.969001e-98## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only## [1] " calling mypredict_mdl for OOB:"
## [1] "in MFO.Classifier$prob"
## N Y
## 1 0.5370177 0.4629823
## 2 0.5370177 0.4629823
## 3 0.5370177 0.4629823
## 4 0.5370177 0.4629823
## 5 0.5370177 0.4629823
## 6 0.5370177 0.4629823
## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.MFO.myMFO_classfr.N
## 1 N 484
## 2 Y 416
## Prediction
## Reference N Y
## N 484 0
## Y 416 0
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 5.377778e-01 0.000000e+00 5.045709e-01 5.707366e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 5.136679e-01 4.930786e-92## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only## id feats max.nTuningRuns min.elapsedtime.everything
## 1 MFO.myMFO_classfr .rnorm 0 0.265
## min.elapsedtime.final max.auc.fit opt.prob.threshold.fit max.f.score.fit
## 1 0.003 0.5 0.5 0
## max.Accuracy.fit max.AccuracyLower.fit max.AccuracyUpper.fit
## 1 0.5370177 0.5048652 0.5689423
## max.Kappa.fit .fit max.auc.OOB opt.prob.threshold.OOB
## 1 0 0.5370177 0.5 0.5
## max.f.score.OOB max.Accuracy.OOB max.AccuracyLower.OOB
## 1 0 0.5377778 0.5045709
## max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.5707366 0 0.5377778if (glb_is_classification)
# "random" model - only for classification;
# none needed for regression since it is same as MFO
ret_lst <- myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst = list(
id.prefix = "Random", type = glb_model_type, trainControl.method = "none",
train.method = "myrandom_classfr")),
indep_vars = ".rnorm", rsp_var = glb_rsp_var,
fit_df = glb_fitobs_df, OOB_df = glb_OOBobs_df)## [1] "fitting model: Random.myrandom_classfr"
## [1] " indep_vars: .rnorm"## Warning in if (mdl_specs_lst[["train.method"]] %in% c("bayesglm", "glm"))
## allowPar <- FALSE: the condition has length > 1 and only the first element
## will be used## Fitting parameter = none on full training set
## Length Class Mode
## unique.vals 2 factor numeric
## unique.prob 2 table numeric
## xNames 1 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 2 -none- character
## [1] " calling mypredict_mdl for fit:"
## [1] "in Random.Classifier$prob"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6329294
## 3 0.2 0.6329294
## 4 0.3 0.6329294
## 5 0.4 0.6329294
## 6 0.5 0.4608501
## 7 0.6 0.0000000
## 8 0.7 0.0000000
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Random.myrandom_classfr.Y
## 1 N 515
## 2 Y 444
## Prediction
## Reference N Y
## N 0 515
## Y 0 444
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 4.629823e-01 0.000000e+00 4.310577e-01 4.951348e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 9.999981e-01 1.409554e-113## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## [1] "in Random.Classifier$prob"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6322188
## 3 0.2 0.6322188
## 4 0.3 0.6322188
## 5 0.4 0.6322188
## 6 0.5 0.4801921
## 7 0.6 0.0000000
## 8 0.7 0.0000000
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Random.myrandom_classfr.Y
## 1 N 484
## 2 Y 416
## Prediction
## Reference N Y
## N 0 484
## Y 0 416
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 4.622222e-01 0.000000e+00 4.292634e-01 4.954291e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 9.999976e-01 7.836108e-107## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Random.myrandom_classfr .rnorm 0
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 0.253 0.002 0.4950888
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.6329294 0.4629823
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.4310577 0.4951348 0 0.4629823
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.5162111 0.4 0.6322188 0.4622222
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.4292634 0.4954291 0 0.4622222# ret_lst <- myfit_mdl(mdl_id = "Random", model_method = "myrandom_classfr",
# model_type = glb_model_type,
# indep_vars_vctr = ".rnorm",
# rsp_var = glb_rsp_var, rsp_var_out = glb_rsp_var_out,
# fit_df = glb_fitobs_df, OOB_df = glb_OOBobs_df)
# Any models that have tuning parameters has "better" results with cross-validation
# (except bag & rf) & "different" results for different outcome metrics
# Max.cor.Y
# Check impact of cv
# rpart is not a good candidate since caret does not optimize cp (only tuning parameter of rpart) well
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix="Max.cor.Y.rcv.1X1", type=glb_model_type, trainControl.method="none",
train.method="glmnet")),
indep_vars=max_cor_y_x_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)## [1] "fitting model: Max.cor.Y.rcv.1X1.glmnet"
## [1] " indep_vars: biddable,startprice"## Loading required package: glmnet
## Loading required package: Matrix
## Loaded glmnet 2.0-2## Fitting alpha = 0.1, lambda = 0.00543 on full training set
## Length Class Mode
## a0 78 -none- numeric
## beta 156 dgCMatrix S4
## df 78 -none- numeric
## dim 2 -none- numeric
## lambda 78 -none- numeric
## dev.ratio 78 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## 0.164304024 1.815169774 -0.005789624
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## 0.167441051 1.819499338 -0.005816932
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6646059
## 3 0.2 0.7144105
## 4 0.3 0.7417747
## 5 0.4 0.7714286
## 6 0.5 0.7549824
## 7 0.6 0.7439024
## 8 0.7 0.7431551
## 9 0.8 0.6912181
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.1X1.glmnet.N
## 1 N 400
## 2 Y 93
## sold.fctr.predict.Max.cor.Y.rcv.1X1.glmnet.Y
## 1 115
## 2 351
## Prediction
## Reference N Y
## N 400 115
## Y 93 351
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.831074e-01 5.653087e-01 7.556528e-01 8.088110e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 1.325363e-56 1.453683e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6563518
## 3 0.2 0.6899724
## 4 0.3 0.7201690
## 5 0.4 0.7473561
## 6 0.5 0.7474747
## 7 0.6 0.7421875
## 8 0.7 0.7140903
## 9 0.8 0.6554364
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.1X1.glmnet.N
## 1 N 404
## 2 Y 120
## sold.fctr.predict.Max.cor.Y.rcv.1X1.glmnet.Y
## 1 80
## 2 296
## Prediction
## Reference N Y
## N 404 80
## Y 120 296
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.777778e-01 5.499640e-01 7.491748e-01 8.045491e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.303387e-50 5.820666e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.1X1.glmnet biddable,startprice 0
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 0.65 0.02 0.8570235
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7714286 0.7831074
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7556528 0.808811 0.5653087 0.7831074
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8253362 0.5 0.7474747 0.7777778
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7491748 0.8045491 0.549964 0.7777778# rcv_n_folds == 1 & rcv_n_repeats > 1 crashes
for (rcv_n_folds in seq(3, glb_rcv_n_folds + 2, 2))
for (rcv_n_repeats in seq(1, glb_rcv_n_repeats + 2, 2)) {
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix=paste0("Max.cor.Y.rcv.", rcv_n_folds, "X", rcv_n_repeats),
type=glb_model_type, trainControl.method="repeatedcv",
trainControl.number=rcv_n_folds, trainControl.repeats=rcv_n_repeats,
train.method="glmnet")),
indep_vars=max_cor_y_x_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)
}## [1] "fitting model: Max.cor.Y.rcv.3X1.glmnet"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 1, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: alpha## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 54 -none- numeric
## beta 108 dgCMatrix S4
## df 54 -none- numeric
## dim 2 -none- numeric
## lambda 54 -none- numeric
## dev.ratio 54 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.177882622 0.988626419 -0.002047359
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.159987259 1.053567455 -0.002286253
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6329294
## 3 0.2 0.6419214
## 4 0.3 0.6912972
## 5 0.4 0.7600849
## 6 0.5 0.7576471
## 7 0.6 0.7431551
## 8 0.7 0.0000000
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X1.glmnet.N
## 1 N 375
## 2 Y 86
## sold.fctr.predict.Max.cor.Y.rcv.3X1.glmnet.Y
## 1 140
## 2 358
## Prediction
## Reference N Y
## N 375 140
## Y 86 358
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.643379e-01 5.300182e-01 7.361668e-01 7.908752e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 3.860897e-48 4.226721e-04## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6326996
## 3 0.2 0.6387597
## 4 0.3 0.6678141
## 5 0.4 0.7350620
## 6 0.5 0.7568922
## 7 0.6 0.7140903
## 8 0.7 0.0000000
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X1.glmnet.N
## 1 N 404
## 2 Y 114
## sold.fctr.predict.Max.cor.Y.rcv.3X1.glmnet.Y
## 1 80
## 2 302
## Prediction
## Reference N Y
## N 404 80
## Y 114 302
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.844444e-01 5.639099e-01 7.561144e-01 8.108984e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.553108e-53 1.782363e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.3X1.glmnet biddable,startprice 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 1.367 0.017 0.8560789
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7600849 0.7883033
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7361668 0.7908752 0.5717602 0.7643379
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8266524 0.5 0.7568922 0.7844444
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7561144 0.8108984 0.5639099 0.7844444
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01755885 0.03350704
## [1] "fitting model: Max.cor.Y.rcv.3X3.glmnet"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.1, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: alpha## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 78 -none- numeric
## beta 156 dgCMatrix S4
## df 78 -none- numeric
## dim 2 -none- numeric
## lambda 78 -none- numeric
## dev.ratio 78 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.05041454 1.20160751 -0.00318479
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.044399804 1.239783450 -0.003306171
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6388489
## 3 0.2 0.6640986
## 4 0.3 0.7220733
## 5 0.4 0.7647691
## 6 0.5 0.7603306
## 7 0.6 0.7450000
## 8 0.7 0.6912181
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X3.glmnet.N
## 1 N 384
## 2 Y 88
## sold.fctr.predict.Max.cor.Y.rcv.3X3.glmnet.Y
## 1 131
## 2 356
## Prediction
## Reference N Y
## N 384 131
## Y 88 356
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.716371e-01 5.438025e-01 7.437357e-01 7.978592e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 2.476899e-51 4.538339e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6369916
## 3 0.2 0.6563518
## 4 0.3 0.6977186
## 5 0.4 0.7382857
## 6 0.5 0.7522013
## 7 0.6 0.7272727
## 8 0.7 0.6554364
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X3.glmnet.N
## 1 N 404
## 2 Y 117
## sold.fctr.predict.Max.cor.Y.rcv.3X3.glmnet.Y
## 1 80
## 2 299
## Prediction
## Reference N Y
## N 404 80
## Y 117 299
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.811111e-01 5.569405e-01 7.526434e-01 8.077250e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 4.632073e-52 1.032074e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.3X3.glmnet biddable,startprice 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 1.834 0.016 0.8568573
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7647691 0.7876154
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7437357 0.7978592 0.5701499 0.7716371
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8260961 0.5 0.7522013 0.7811111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7526434 0.807725 0.5569405 0.7811111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01255994 0.02551235
## [1] "fitting model: Max.cor.Y.rcv.3X5.glmnet"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.1, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: alpha## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 78 -none- numeric
## beta 156 dgCMatrix S4
## df 78 -none- numeric
## dim 2 -none- numeric
## lambda 78 -none- numeric
## dev.ratio 78 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.05041454 1.20160751 -0.00318479
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.044399804 1.239783450 -0.003306171
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6388489
## 3 0.2 0.6640986
## 4 0.3 0.7220733
## 5 0.4 0.7647691
## 6 0.5 0.7603306
## 7 0.6 0.7450000
## 8 0.7 0.6912181
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X5.glmnet.N
## 1 N 384
## 2 Y 88
## sold.fctr.predict.Max.cor.Y.rcv.3X5.glmnet.Y
## 1 131
## 2 356
## Prediction
## Reference N Y
## N 384 131
## Y 88 356
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.716371e-01 5.438025e-01 7.437357e-01 7.978592e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 2.476899e-51 4.538339e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6369916
## 3 0.2 0.6563518
## 4 0.3 0.6977186
## 5 0.4 0.7382857
## 6 0.5 0.7522013
## 7 0.6 0.7272727
## 8 0.7 0.6554364
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X5.glmnet.N
## 1 N 404
## 2 Y 117
## sold.fctr.predict.Max.cor.Y.rcv.3X5.glmnet.Y
## 1 80
## 2 299
## Prediction
## Reference N Y
## N 404 80
## Y 117 299
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.811111e-01 5.569405e-01 7.526434e-01 8.077250e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 4.632073e-52 1.032074e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.3X5.glmnet biddable,startprice 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 2.335 0.017 0.8568573
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7647691 0.7879043
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7437357 0.7978592 0.5707133 0.7716371
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8260961 0.5 0.7522013 0.7811111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7526434 0.807725 0.5569405 0.7811111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.016552 0.03392463
## [1] "fitting model: Max.cor.Y.rcv.5X1.glmnet"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 1, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: alpha## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 54 -none- numeric
## beta 108 dgCMatrix S4
## df 54 -none- numeric
## dim 2 -none- numeric
## lambda 54 -none- numeric
## dev.ratio 54 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.177882622 0.988626419 -0.002047359
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.159987259 1.053567455 -0.002286253
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6329294
## 3 0.2 0.6419214
## 4 0.3 0.6912972
## 5 0.4 0.7600849
## 6 0.5 0.7576471
## 7 0.6 0.7431551
## 8 0.7 0.0000000
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.5X1.glmnet.N
## 1 N 375
## 2 Y 86
## sold.fctr.predict.Max.cor.Y.rcv.5X1.glmnet.Y
## 1 140
## 2 358
## Prediction
## Reference N Y
## N 375 140
## Y 86 358
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.643379e-01 5.300182e-01 7.361668e-01 7.908752e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 3.860897e-48 4.226721e-04## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6326996
## 3 0.2 0.6387597
## 4 0.3 0.6678141
## 5 0.4 0.7350620
## 6 0.5 0.7568922
## 7 0.6 0.7140903
## 8 0.7 0.0000000
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.5X1.glmnet.N
## 1 N 404
## 2 Y 114
## sold.fctr.predict.Max.cor.Y.rcv.5X1.glmnet.Y
## 1 80
## 2 302
## Prediction
## Reference N Y
## N 404 80
## Y 114 302
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.844444e-01 5.639099e-01 7.561144e-01 8.108984e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.553108e-53 1.782363e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.5X1.glmnet biddable,startprice 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 1.558 0.014 0.8560789
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7600849 0.7872982
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7361668 0.7908752 0.5698703 0.7643379
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8266524 0.5 0.7568922 0.7844444
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7561144 0.8108984 0.5639099 0.7844444
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01714756 0.03415616
## [1] "fitting model: Max.cor.Y.rcv.5X3.glmnet"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.1, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: alpha## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 78 -none- numeric
## beta 156 dgCMatrix S4
## df 78 -none- numeric
## dim 2 -none- numeric
## lambda 78 -none- numeric
## dev.ratio 78 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.05041454 1.20160751 -0.00318479
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.044399804 1.239783450 -0.003306171
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6388489
## 3 0.2 0.6640986
## 4 0.3 0.7220733
## 5 0.4 0.7647691
## 6 0.5 0.7603306
## 7 0.6 0.7450000
## 8 0.7 0.6912181
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.5X3.glmnet.N
## 1 N 384
## 2 Y 88
## sold.fctr.predict.Max.cor.Y.rcv.5X3.glmnet.Y
## 1 131
## 2 356
## Prediction
## Reference N Y
## N 384 131
## Y 88 356
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.716371e-01 5.438025e-01 7.437357e-01 7.978592e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 2.476899e-51 4.538339e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6369916
## 3 0.2 0.6563518
## 4 0.3 0.6977186
## 5 0.4 0.7382857
## 6 0.5 0.7522013
## 7 0.6 0.7272727
## 8 0.7 0.6554364
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.5X3.glmnet.N
## 1 N 404
## 2 Y 117
## sold.fctr.predict.Max.cor.Y.rcv.5X3.glmnet.Y
## 1 80
## 2 299
## Prediction
## Reference N Y
## N 404 80
## Y 117 299
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.811111e-01 5.569405e-01 7.526434e-01 8.077250e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 4.632073e-52 1.032074e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.5X3.glmnet biddable,startprice 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 2.252 0.017 0.8568573
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7647691 0.7879781
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7437357 0.7978592 0.5707928 0.7716371
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8260961 0.5 0.7522013 0.7811111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7526434 0.807725 0.5569405 0.7811111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.02067742 0.04250265
## [1] "fitting model: Max.cor.Y.rcv.5X5.glmnet"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.1, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: alpha## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = paste0("Max.cor.Y.rcv.", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 78 -none- numeric
## beta 156 dgCMatrix S4
## df 78 -none- numeric
## dim 2 -none- numeric
## lambda 78 -none- numeric
## dev.ratio 78 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 2 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.05041454 1.20160751 -0.00318479
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.044399804 1.239783450 -0.003306171
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6388489
## 3 0.2 0.6640986
## 4 0.3 0.7220733
## 5 0.4 0.7647691
## 6 0.5 0.7603306
## 7 0.6 0.7450000
## 8 0.7 0.6912181
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.5X5.glmnet.N
## 1 N 384
## 2 Y 88
## sold.fctr.predict.Max.cor.Y.rcv.5X5.glmnet.Y
## 1 131
## 2 356
## Prediction
## Reference N Y
## N 384 131
## Y 88 356
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.716371e-01 5.438025e-01 7.437357e-01 7.978592e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 2.476899e-51 4.538339e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6369916
## 3 0.2 0.6563518
## 4 0.3 0.6977186
## 5 0.4 0.7382857
## 6 0.5 0.7522013
## 7 0.6 0.7272727
## 8 0.7 0.6554364
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.5X5.glmnet.N
## 1 N 404
## 2 Y 117
## sold.fctr.predict.Max.cor.Y.rcv.5X5.glmnet.Y
## 1 80
## 2 299
## Prediction
## Reference N Y
## N 404 80
## Y 117 299
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.811111e-01 5.569405e-01 7.526434e-01 8.077250e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 4.632073e-52 1.032074e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.5X5.glmnet biddable,startprice 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 2.869 0.016 0.8568573
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7647691 0.7874989
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7437357 0.7978592 0.5698279 0.7716371
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8260961 0.5 0.7522013 0.7811111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7526434 0.807725 0.5569405 0.7811111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.02952906 0.06009372# Add parallel coordinates graph of glb_models_df[, glb_model_evl_criteria] to evaluate cv parameters
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix="Max.cor.Y.rcv.1X1.cp.0", type=glb_model_type, trainControl.method="none",
train.method="rpart",
tune.df=data.frame(method="rpart", parameter="cp", min=0.0, max=0.0, by=0.1))),
indep_vars=max_cor_y_x_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)## [1] "fitting model: Max.cor.Y.rcv.1X1.cp.0.rpart"
## [1] " indep_vars: biddable,startprice"## Loading required package: rpart## Fitting cp = 0 on full training set## Loading required package: rpart.plot
## Call:
## rpart(formula = .outcome ~ ., control = list(minsplit = 20, minbucket = 7,
## cp = 0, maxcompete = 4, maxsurrogate = 5, usesurrogate = 2,
## surrogatestyle = 0, maxdepth = 30, xval = 0))
## n= 959
##
## CP nsplit rel error
## 1 0.5112612613 0 1.0000000
## 2 0.0270270270 1 0.4887387
## 3 0.0056306306 3 0.4346847
## 4 0.0045045045 5 0.4234234
## 5 0.0030030030 10 0.4009009
## 6 0.0015015015 14 0.3851351
## 7 0.0011261261 20 0.3761261
## 8 0.0007507508 22 0.3738739
## 9 0.0003753754 25 0.3716216
## 10 0.0000000000 31 0.3693694
##
## Variable importance
## startprice biddable
## 53 47
##
## Node number 1: 959 observations, complexity param=0.5112613
## predicted class=N expected loss=0.4629823 P(node) =1
## class counts: 515 444
## probabilities: 0.537 0.463
## left son=2 (526 obs) right son=3 (433 obs)
## Primary splits:
## biddable < 0.5 to the left, improve=141.2885, (0 missing)
## startprice < 100.5 to the right, improve=136.2337, (0 missing)
## Surrogate splits:
## startprice < 100.5 to the right, agree=0.755, adj=0.457, (0 split)
##
## Node number 2: 526 observations, complexity param=0.005630631
## predicted class=N expected loss=0.21673 P(node) =0.548488
## class counts: 412 114
## probabilities: 0.783 0.217
## left son=4 (474 obs) right son=5 (52 obs)
## Primary splits:
## startprice < 93 to the right, improve=9.260657, (0 missing)
##
## Node number 3: 433 observations, complexity param=0.02702703
## predicted class=Y expected loss=0.2378753 P(node) =0.451512
## class counts: 103 330
## probabilities: 0.238 0.762
## left son=6 (153 obs) right son=7 (280 obs)
## Primary splits:
## startprice < 129.995 to the right, improve=38.43639, (0 missing)
##
## Node number 4: 474 observations, complexity param=0.004504505
## predicted class=N expected loss=0.185654 P(node) =0.4942649
## class counts: 386 88
## probabilities: 0.814 0.186
## left son=8 (271 obs) right son=9 (203 obs)
## Primary splits:
## startprice < 249.995 to the right, improve=3.529864, (0 missing)
##
## Node number 5: 52 observations, complexity param=0.005630631
## predicted class=N expected loss=0.5 P(node) =0.05422315
## class counts: 26 26
## probabilities: 0.500 0.500
## left son=10 (9 obs) right son=11 (43 obs)
## Primary splits:
## startprice < 28.975 to the left, improve=1.679587, (0 missing)
##
## Node number 6: 153 observations, complexity param=0.02702703
## predicted class=N expected loss=0.4771242 P(node) =0.1595412
## class counts: 80 73
## probabilities: 0.523 0.477
## left son=12 (92 obs) right son=13 (61 obs)
## Primary splits:
## startprice < 205 to the right, improve=5.339157, (0 missing)
##
## Node number 7: 280 observations, complexity param=0.0007507508
## predicted class=Y expected loss=0.08214286 P(node) =0.2919708
## class counts: 23 257
## probabilities: 0.082 0.918
## left son=14 (63 obs) right son=15 (217 obs)
## Primary splits:
## startprice < 93.74 to the right, improve=3.954147, (0 missing)
##
## Node number 8: 271 observations, complexity param=0.004504505
## predicted class=N expected loss=0.1328413 P(node) =0.282586
## class counts: 235 36
## probabilities: 0.867 0.133
## left son=16 (98 obs) right son=17 (173 obs)
## Primary splits:
## startprice < 424.995 to the right, improve=0.8051306, (0 missing)
##
## Node number 9: 203 observations, complexity param=0.004504505
## predicted class=N expected loss=0.2561576 P(node) =0.2116788
## class counts: 151 52
## probabilities: 0.744 0.256
## left son=18 (101 obs) right son=19 (102 obs)
## Primary splits:
## startprice < 175.635 to the left, improve=1.358829, (0 missing)
##
## Node number 10: 9 observations
## predicted class=N expected loss=0.2222222 P(node) =0.009384776
## class counts: 7 2
## probabilities: 0.778 0.222
##
## Node number 11: 43 observations, complexity param=0.001501502
## predicted class=Y expected loss=0.4418605 P(node) =0.04483837
## class counts: 19 24
## probabilities: 0.442 0.558
## left son=22 (31 obs) right son=23 (12 obs)
## Primary splits:
## startprice < 59.5 to the right, improve=0.392098, (0 missing)
##
## Node number 12: 92 observations, complexity param=0.003003003
## predicted class=N expected loss=0.3695652 P(node) =0.09593326
## class counts: 58 34
## probabilities: 0.630 0.370
## left son=24 (17 obs) right son=25 (75 obs)
## Primary splits:
## startprice < 240.995 to the left, improve=2.64682, (0 missing)
##
## Node number 13: 61 observations, complexity param=0.001126126
## predicted class=Y expected loss=0.3606557 P(node) =0.06360792
## class counts: 22 39
## probabilities: 0.361 0.639
## left son=26 (27 obs) right son=27 (34 obs)
## Primary splits:
## startprice < 157.495 to the left, improve=1.414372, (0 missing)
##
## Node number 14: 63 observations, complexity param=0.0007507508
## predicted class=Y expected loss=0.2380952 P(node) =0.06569343
## class counts: 15 48
## probabilities: 0.238 0.762
## left son=28 (52 obs) right son=29 (11 obs)
## Primary splits:
## startprice < 122.5 to the left, improve=0.5774226, (0 missing)
##
## Node number 15: 217 observations
## predicted class=Y expected loss=0.03686636 P(node) =0.2262774
## class counts: 8 209
## probabilities: 0.037 0.963
##
## Node number 16: 98 observations
## predicted class=N expected loss=0.08163265 P(node) =0.1021898
## class counts: 90 8
## probabilities: 0.918 0.082
##
## Node number 17: 173 observations, complexity param=0.004504505
## predicted class=N expected loss=0.1618497 P(node) =0.1803962
## class counts: 145 28
## probabilities: 0.838 0.162
## left son=34 (166 obs) right son=35 (7 obs)
## Primary splits:
## startprice < 404 to the left, improve=7.053456, (0 missing)
##
## Node number 18: 101 observations, complexity param=0.001501502
## predicted class=N expected loss=0.1980198 P(node) =0.105318
## class counts: 81 20
## probabilities: 0.802 0.198
## left son=36 (38 obs) right son=37 (63 obs)
## Primary splits:
## startprice < 152.43 to the right, improve=1.727495, (0 missing)
##
## Node number 19: 102 observations, complexity param=0.004504505
## predicted class=N expected loss=0.3137255 P(node) =0.1063608
## class counts: 70 32
## probabilities: 0.686 0.314
## left son=38 (89 obs) right son=39 (13 obs)
## Primary splits:
## startprice < 185.245 to the right, improve=4.270748, (0 missing)
##
## Node number 22: 31 observations, complexity param=0.001501502
## predicted class=Y expected loss=0.483871 P(node) =0.03232534
## class counts: 15 16
## probabilities: 0.484 0.516
## left son=44 (7 obs) right son=45 (24 obs)
## Primary splits:
## startprice < 74.975 to the left, improve=0.1386329, (0 missing)
##
## Node number 23: 12 observations
## predicted class=Y expected loss=0.3333333 P(node) =0.01251303
## class counts: 4 8
## probabilities: 0.333 0.667
##
## Node number 24: 17 observations
## predicted class=N expected loss=0.1176471 P(node) =0.0177268
## class counts: 15 2
## probabilities: 0.882 0.118
##
## Node number 25: 75 observations, complexity param=0.003003003
## predicted class=N expected loss=0.4266667 P(node) =0.07820647
## class counts: 43 32
## probabilities: 0.573 0.427
## left son=50 (15 obs) right son=51 (60 obs)
## Primary splits:
## startprice < 494.995 to the right, improve=3.226667, (0 missing)
##
## Node number 26: 27 observations, complexity param=0.001126126
## predicted class=Y expected loss=0.4814815 P(node) =0.02815433
## class counts: 13 14
## probabilities: 0.481 0.519
## left son=52 (17 obs) right son=53 (10 obs)
## Primary splits:
## startprice < 149.995 to the right, improve=0.2108932, (0 missing)
##
## Node number 27: 34 observations
## predicted class=Y expected loss=0.2647059 P(node) =0.0354536
## class counts: 9 25
## probabilities: 0.265 0.735
##
## Node number 28: 52 observations, complexity param=0.0007507508
## predicted class=Y expected loss=0.2692308 P(node) =0.05422315
## class counts: 14 38
## probabilities: 0.269 0.731
## left son=56 (7 obs) right son=57 (45 obs)
## Primary splits:
## startprice < 105 to the right, improve=1.477411, (0 missing)
##
## Node number 29: 11 observations
## predicted class=Y expected loss=0.09090909 P(node) =0.01147028
## class counts: 1 10
## probabilities: 0.091 0.909
##
## Node number 34: 166 observations
## predicted class=N expected loss=0.1325301 P(node) =0.173097
## class counts: 144 22
## probabilities: 0.867 0.133
##
## Node number 35: 7 observations
## predicted class=Y expected loss=0.1428571 P(node) =0.00729927
## class counts: 1 6
## probabilities: 0.143 0.857
##
## Node number 36: 38 observations
## predicted class=N expected loss=0.07894737 P(node) =0.03962461
## class counts: 35 3
## probabilities: 0.921 0.079
##
## Node number 37: 63 observations, complexity param=0.001501502
## predicted class=N expected loss=0.2698413 P(node) =0.06569343
## class counts: 46 17
## probabilities: 0.730 0.270
## left son=74 (39 obs) right son=75 (24 obs)
## Primary splits:
## startprice < 129.995 to the left, improve=2.754884, (0 missing)
##
## Node number 38: 89 observations, complexity param=0.0003753754
## predicted class=N expected loss=0.258427 P(node) =0.09280501
## class counts: 66 23
## probabilities: 0.742 0.258
## left son=76 (9 obs) right son=77 (80 obs)
## Primary splits:
## startprice < 189.495 to the left, improve=1.33736, (0 missing)
##
## Node number 39: 13 observations
## predicted class=Y expected loss=0.3076923 P(node) =0.01355579
## class counts: 4 9
## probabilities: 0.308 0.692
##
## Node number 44: 7 observations
## predicted class=N expected loss=0.4285714 P(node) =0.00729927
## class counts: 4 3
## probabilities: 0.571 0.429
##
## Node number 45: 24 observations, complexity param=0.001501502
## predicted class=Y expected loss=0.4583333 P(node) =0.02502607
## class counts: 11 13
## probabilities: 0.458 0.542
## left son=90 (7 obs) right son=91 (17 obs)
## Primary splits:
## startprice < 89.25 to the right, improve=0.2528011, (0 missing)
##
## Node number 50: 15 observations
## predicted class=N expected loss=0.1333333 P(node) =0.01564129
## class counts: 13 2
## probabilities: 0.867 0.133
##
## Node number 51: 60 observations, complexity param=0.003003003
## predicted class=N expected loss=0.5 P(node) =0.06256517
## class counts: 30 30
## probabilities: 0.500 0.500
## left son=102 (18 obs) right son=103 (42 obs)
## Primary splits:
## startprice < 270 to the left, improve=0.6349206, (0 missing)
##
## Node number 52: 17 observations
## predicted class=N expected loss=0.4705882 P(node) =0.0177268
## class counts: 9 8
## probabilities: 0.529 0.471
##
## Node number 53: 10 observations
## predicted class=Y expected loss=0.4 P(node) =0.01042753
## class counts: 4 6
## probabilities: 0.400 0.600
##
## Node number 56: 7 observations
## predicted class=N expected loss=0.4285714 P(node) =0.00729927
## class counts: 4 3
## probabilities: 0.571 0.429
##
## Node number 57: 45 observations
## predicted class=Y expected loss=0.2222222 P(node) =0.04692388
## class counts: 10 35
## probabilities: 0.222 0.778
##
## Node number 74: 39 observations
## predicted class=N expected loss=0.1538462 P(node) =0.04066736
## class counts: 33 6
## probabilities: 0.846 0.154
##
## Node number 75: 24 observations, complexity param=0.001501502
## predicted class=N expected loss=0.4583333 P(node) =0.02502607
## class counts: 13 11
## probabilities: 0.542 0.458
## left son=150 (14 obs) right son=151 (10 obs)
## Primary splits:
## startprice < 149.995 to the left, improve=0.6880952, (0 missing)
##
## Node number 76: 9 observations
## predicted class=N expected loss=0 P(node) =0.009384776
## class counts: 9 0
## probabilities: 1.000 0.000
##
## Node number 77: 80 observations, complexity param=0.0003753754
## predicted class=N expected loss=0.2875 P(node) =0.08342023
## class counts: 57 23
## probabilities: 0.713 0.287
## left son=154 (71 obs) right son=155 (9 obs)
## Primary splits:
## startprice < 240.94 to the left, improve=0.4995696, (0 missing)
##
## Node number 90: 7 observations
## predicted class=N expected loss=0.4285714 P(node) =0.00729927
## class counts: 4 3
## probabilities: 0.571 0.429
##
## Node number 91: 17 observations
## predicted class=Y expected loss=0.4117647 P(node) =0.0177268
## class counts: 7 10
## probabilities: 0.412 0.588
##
## Node number 102: 18 observations
## predicted class=N expected loss=0.3888889 P(node) =0.01876955
## class counts: 11 7
## probabilities: 0.611 0.389
##
## Node number 103: 42 observations, complexity param=0.003003003
## predicted class=Y expected loss=0.452381 P(node) =0.04379562
## class counts: 19 23
## probabilities: 0.452 0.548
## left son=206 (21 obs) right son=207 (21 obs)
## Primary splits:
## startprice < 319.995 to the right, improve=1.190476, (0 missing)
##
## Node number 150: 14 observations
## predicted class=N expected loss=0.3571429 P(node) =0.01459854
## class counts: 9 5
## probabilities: 0.643 0.357
##
## Node number 151: 10 observations
## predicted class=Y expected loss=0.4 P(node) =0.01042753
## class counts: 4 6
## probabilities: 0.400 0.600
##
## Node number 154: 71 observations, complexity param=0.0003753754
## predicted class=N expected loss=0.2676056 P(node) =0.07403545
## class counts: 52 19
## probabilities: 0.732 0.268
## left son=308 (9 obs) right son=309 (62 obs)
## Primary splits:
## startprice < 229.995 to the right, improve=0.504821, (0 missing)
##
## Node number 155: 9 observations
## predicted class=N expected loss=0.4444444 P(node) =0.009384776
## class counts: 5 4
## probabilities: 0.556 0.444
##
## Node number 206: 21 observations
## predicted class=N expected loss=0.4285714 P(node) =0.02189781
## class counts: 12 9
## probabilities: 0.571 0.429
##
## Node number 207: 21 observations
## predicted class=Y expected loss=0.3333333 P(node) =0.02189781
## class counts: 7 14
## probabilities: 0.333 0.667
##
## Node number 308: 9 observations
## predicted class=N expected loss=0.1111111 P(node) =0.009384776
## class counts: 8 1
## probabilities: 0.889 0.111
##
## Node number 309: 62 observations, complexity param=0.0003753754
## predicted class=N expected loss=0.2903226 P(node) =0.06465068
## class counts: 44 18
## probabilities: 0.710 0.290
## left son=618 (55 obs) right son=619 (7 obs)
## Primary splits:
## startprice < 226.44 to the left, improve=0.3016339, (0 missing)
##
## Node number 618: 55 observations, complexity param=0.0003753754
## predicted class=N expected loss=0.2727273 P(node) =0.05735141
## class counts: 40 15
## probabilities: 0.727 0.273
## left son=1236 (10 obs) right son=1237 (45 obs)
## Primary splits:
## startprice < 219.995 to the right, improve=0.7292929, (0 missing)
##
## Node number 619: 7 observations
## predicted class=N expected loss=0.4285714 P(node) =0.00729927
## class counts: 4 3
## probabilities: 0.571 0.429
##
## Node number 1236: 10 observations
## predicted class=N expected loss=0.1 P(node) =0.01042753
## class counts: 9 1
## probabilities: 0.900 0.100
##
## Node number 1237: 45 observations, complexity param=0.0003753754
## predicted class=N expected loss=0.3111111 P(node) =0.04692388
## class counts: 31 14
## probabilities: 0.689 0.311
## left son=2474 (38 obs) right son=2475 (7 obs)
## Primary splits:
## startprice < 214.995 to the left, improve=1.123475, (0 missing)
##
## Node number 2474: 38 observations
## predicted class=N expected loss=0.2631579 P(node) =0.03962461
## class counts: 28 10
## probabilities: 0.737 0.263
##
## Node number 2475: 7 observations
## predicted class=Y expected loss=0.4285714 P(node) =0.00729927
## class counts: 3 4
## probabilities: 0.429 0.571
##
## n= 959
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 959 444 N (0.53701773 0.46298227)
## 2) biddable< 0.5 526 114 N (0.78326996 0.21673004)
## 4) startprice>=93 474 88 N (0.81434599 0.18565401)
## 8) startprice>=249.995 271 36 N (0.86715867 0.13284133)
## 16) startprice>=424.995 98 8 N (0.91836735 0.08163265) *
## 17) startprice< 424.995 173 28 N (0.83815029 0.16184971)
## 34) startprice< 404 166 22 N (0.86746988 0.13253012) *
## 35) startprice>=404 7 1 Y (0.14285714 0.85714286) *
## 9) startprice< 249.995 203 52 N (0.74384236 0.25615764)
## 18) startprice< 175.635 101 20 N (0.80198020 0.19801980)
## 36) startprice>=152.43 38 3 N (0.92105263 0.07894737) *
## 37) startprice< 152.43 63 17 N (0.73015873 0.26984127)
## 74) startprice< 129.995 39 6 N (0.84615385 0.15384615) *
## 75) startprice>=129.995 24 11 N (0.54166667 0.45833333)
## 150) startprice< 149.995 14 5 N (0.64285714 0.35714286) *
## 151) startprice>=149.995 10 4 Y (0.40000000 0.60000000) *
## 19) startprice>=175.635 102 32 N (0.68627451 0.31372549)
## 38) startprice>=185.245 89 23 N (0.74157303 0.25842697)
## 76) startprice< 189.495 9 0 N (1.00000000 0.00000000) *
## 77) startprice>=189.495 80 23 N (0.71250000 0.28750000)
## 154) startprice< 240.94 71 19 N (0.73239437 0.26760563)
## 308) startprice>=229.995 9 1 N (0.88888889 0.11111111) *
## 309) startprice< 229.995 62 18 N (0.70967742 0.29032258)
## 618) startprice< 226.44 55 15 N (0.72727273 0.27272727)
## 1236) startprice>=219.995 10 1 N (0.90000000 0.10000000) *
## 1237) startprice< 219.995 45 14 N (0.68888889 0.31111111)
## 2474) startprice< 214.995 38 10 N (0.73684211 0.26315789) *
## 2475) startprice>=214.995 7 3 Y (0.42857143 0.57142857) *
## 619) startprice>=226.44 7 3 N (0.57142857 0.42857143) *
## 155) startprice>=240.94 9 4 N (0.55555556 0.44444444) *
## 39) startprice< 185.245 13 4 Y (0.30769231 0.69230769) *
## 5) startprice< 93 52 26 N (0.50000000 0.50000000)
## 10) startprice< 28.975 9 2 N (0.77777778 0.22222222) *
## 11) startprice>=28.975 43 19 Y (0.44186047 0.55813953)
## 22) startprice>=59.5 31 15 Y (0.48387097 0.51612903)
## 44) startprice< 74.975 7 3 N (0.57142857 0.42857143) *
## 45) startprice>=74.975 24 11 Y (0.45833333 0.54166667)
## 90) startprice>=89.25 7 3 N (0.57142857 0.42857143) *
## 91) startprice< 89.25 17 7 Y (0.41176471 0.58823529) *
## 23) startprice< 59.5 12 4 Y (0.33333333 0.66666667) *
## 3) biddable>=0.5 433 103 Y (0.23787529 0.76212471)
## 6) startprice>=129.995 153 73 N (0.52287582 0.47712418)
## 12) startprice>=205 92 34 N (0.63043478 0.36956522)
## 24) startprice< 240.995 17 2 N (0.88235294 0.11764706) *
## 25) startprice>=240.995 75 32 N (0.57333333 0.42666667)
## 50) startprice>=494.995 15 2 N (0.86666667 0.13333333) *
## 51) startprice< 494.995 60 30 N (0.50000000 0.50000000)
## 102) startprice< 270 18 7 N (0.61111111 0.38888889) *
## 103) startprice>=270 42 19 Y (0.45238095 0.54761905)
## 206) startprice>=319.995 21 9 N (0.57142857 0.42857143) *
## 207) startprice< 319.995 21 7 Y (0.33333333 0.66666667) *
## 13) startprice< 205 61 22 Y (0.36065574 0.63934426)
## 26) startprice< 157.495 27 13 Y (0.48148148 0.51851852)
## 52) startprice>=149.995 17 8 N (0.52941176 0.47058824) *
## 53) startprice< 149.995 10 4 Y (0.40000000 0.60000000) *
## 27) startprice>=157.495 34 9 Y (0.26470588 0.73529412) *
## 7) startprice< 129.995 280 23 Y (0.08214286 0.91785714)
## 14) startprice>=93.74 63 15 Y (0.23809524 0.76190476)
## 28) startprice< 122.5 52 14 Y (0.26923077 0.73076923)
## 56) startprice>=105 7 3 N (0.57142857 0.42857143) *
## 57) startprice< 105 45 10 Y (0.22222222 0.77777778) *
## 29) startprice>=122.5 11 1 Y (0.09090909 0.90909091) *
## 15) startprice< 93.74 217 8 Y (0.03686636 0.96313364) *
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6923077
## 3 0.2 0.7964072
## 4 0.3 0.8104712
## 5 0.4 0.8125677
## 6 0.5 0.8066038
## 7 0.6 0.7860697
## 8 0.7 0.7519789
## 9 0.8 0.6627393
## 10 0.9 0.6517857
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.1X1.cp.0.rpart.N
## 1 N 411
## 2 Y 69
## sold.fctr.predict.Max.cor.Y.rcv.1X1.cp.0.rpart.Y
## 1 104
## 2 375
## Prediction
## Reference N Y
## N 411 104
## Y 69 375
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.196038e-01 6.391797e-01 7.937768e-01 8.434503e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 1.399521e-75 9.738687e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6600000
## 3 0.2 0.7110656
## 4 0.3 0.7260870
## 5 0.4 0.7137970
## 6 0.5 0.6983312
## 7 0.6 0.6937669
## 8 0.7 0.6751773
## 9 0.8 0.5797101
## 10 0.9 0.5798046
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.3000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.1X1.cp.0.rpart.N
## 1 N 314
## 2 Y 82
## sold.fctr.predict.Max.cor.Y.rcv.1X1.cp.0.rpart.Y
## 1 170
## 2 334
## Prediction
## Reference N Y
## N 314 170
## Y 82 334
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.200000e-01 4.450317e-01 6.894226e-01 7.491296e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 2.941697e-29 4.241618e-08## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.1X1.cp.0.rpart biddable,startprice 0
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 0.693 0.012 0.8939014
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.8125677 0.8196038
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7937768 0.8434503 0.6391797 0.8196038
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.7971655 0.3 0.726087 0.72
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.6894226 0.7491296 0.4450317 0.72# if (glb_is_regression || glb_is_binomial) # For multinomials this model will be run next by default
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix="Max.cor.Y",
type=glb_model_type, trainControl.method="repeatedcv",
trainControl.number=glb_rcv_n_folds, trainControl.repeats=glb_rcv_n_repeats,
train.method="rpart")),
indep_vars=max_cor_y_x_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)## [1] "fitting model: Max.cor.Y.rpart"
## [1] " indep_vars: biddable,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting cp = 0.00563 on full training set
## Call:
## rpart(formula = .outcome ~ ., control = list(minsplit = 20, minbucket = 7,
## cp = 0, maxcompete = 4, maxsurrogate = 5, usesurrogate = 2,
## surrogatestyle = 0, maxdepth = 30, xval = 0))
## n= 959
##
## CP nsplit rel error
## 1 0.511261261 0 1.0000000
## 2 0.027027027 1 0.4887387
## 3 0.005630631 3 0.4346847
##
## Variable importance
## biddable startprice
## 57 43
##
## Node number 1: 959 observations, complexity param=0.5112613
## predicted class=N expected loss=0.4629823 P(node) =1
## class counts: 515 444
## probabilities: 0.537 0.463
## left son=2 (526 obs) right son=3 (433 obs)
## Primary splits:
## biddable < 0.5 to the left, improve=141.2885, (0 missing)
## startprice < 100.5 to the right, improve=136.2337, (0 missing)
## Surrogate splits:
## startprice < 100.5 to the right, agree=0.755, adj=0.457, (0 split)
##
## Node number 2: 526 observations
## predicted class=N expected loss=0.21673 P(node) =0.548488
## class counts: 412 114
## probabilities: 0.783 0.217
##
## Node number 3: 433 observations, complexity param=0.02702703
## predicted class=Y expected loss=0.2378753 P(node) =0.451512
## class counts: 103 330
## probabilities: 0.238 0.762
## left son=6 (153 obs) right son=7 (280 obs)
## Primary splits:
## startprice < 129.995 to the right, improve=38.43639, (0 missing)
##
## Node number 6: 153 observations, complexity param=0.02702703
## predicted class=N expected loss=0.4771242 P(node) =0.1595412
## class counts: 80 73
## probabilities: 0.523 0.477
## left son=12 (92 obs) right son=13 (61 obs)
## Primary splits:
## startprice < 205 to the right, improve=5.339157, (0 missing)
##
## Node number 7: 280 observations
## predicted class=Y expected loss=0.08214286 P(node) =0.2919708
## class counts: 23 257
## probabilities: 0.082 0.918
##
## Node number 12: 92 observations
## predicted class=N expected loss=0.3695652 P(node) =0.09593326
## class counts: 58 34
## probabilities: 0.630 0.370
##
## Node number 13: 61 observations
## predicted class=Y expected loss=0.3606557 P(node) =0.06360792
## class counts: 22 39
## probabilities: 0.361 0.639
##
## n= 959
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 959 444 N (0.53701773 0.46298227)
## 2) biddable< 0.5 526 114 N (0.78326996 0.21673004) *
## 3) biddable>=0.5 433 103 Y (0.23787529 0.76212471)
## 6) startprice>=129.995 153 73 N (0.52287582 0.47712418)
## 12) startprice>=205 92 34 N (0.63043478 0.36956522) *
## 13) startprice< 205 61 22 Y (0.36065574 0.63934426) *
## 7) startprice< 129.995 280 23 Y (0.08214286 0.91785714) *
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6329294
## 3 0.2 0.6329294
## 4 0.3 0.7525656
## 5 0.4 0.7541401
## 6 0.5 0.7541401
## 7 0.6 0.7541401
## 8 0.7 0.7099448
## 9 0.8 0.7099448
## 10 0.9 0.7099448
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.6000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rpart.N
## 1 N 470
## 2 Y 148
## sold.fctr.predict.Max.cor.Y.rpart.Y
## 1 45
## 2 296
## Prediction
## Reference N Y
## N 470 45
## Y 148 296
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.987487e-01 5.887020e-01 7.719514e-01 8.236969e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 2.591755e-64 2.102357e-13## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6322188
## 3 0.2 0.6322188
## 4 0.3 0.7560976
## 5 0.4 0.7258065
## 6 0.5 0.7258065
## 7 0.6 0.7258065
## 8 0.7 0.6845238
## 9 0.8 0.6845238
## 10 0.9 0.6845238
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.3000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rpart.N
## 1 N 390
## 2 Y 106
## sold.fctr.predict.Max.cor.Y.rpart.Y
## 1 94
## 2 310
## Prediction
## Reference N Y
## N 390 94
## Y 106 310
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.777778e-01 5.520962e-01 7.491748e-01 8.045491e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.303387e-50 4.366766e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rpart biddable,startprice 5
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 1.605 0.012 0.8162184
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.6 0.7541401 0.7824049
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7719514 0.8236969 0.5580889 0.7987487
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8095598 0.3 0.7560976 0.7777778
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7491748 0.8045491 0.5520962 0.7777778
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01313272 0.02681447if (!is.null(glb_date_vars) &&
(sum(grepl(paste(glb_date_vars, "\\.day\\.minutes\\.poly\\.", sep=""),
names(glb_allobs_df))) > 0)) {
# ret_lst <- myfit_mdl(mdl_id="Max.cor.Y.TmSrs.poly1",
# model_method=ifelse(glb_is_regression, "lm",
# ifelse(glb_is_binomial, "glm", "rpart")),
# model_type=glb_model_type,
# indep_vars_vctr=c(max_cor_y_x_vars, paste0(glb_date_vars, ".day.minutes")),
# rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
# fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df,
# n_cv_folds=glb_rcv_n_folds, tune_models_df=NULL)
#
ret_lst <- myfit_mdl(mdl_id="Max.cor.Y.TmSrs.poly",
model_method=ifelse(glb_is_regression, "lm",
ifelse(glb_is_binomial, "glm", "rpart")),
model_type=glb_model_type,
indep_vars_vctr=c(max_cor_y_x_vars,
grep(paste(glb_date_vars, "\\.day\\.minutes\\.poly\\.", sep=""),
names(glb_allobs_df), value=TRUE)),
rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df,
n_cv_folds=glb_rcv_n_folds, tune_models_df=NULL)
}
# Interactions.High.cor.Y
if (length(int_feats <- setdiff(setdiff(unique(glb_feats_df$cor.high.X), NA),
subset(glb_feats_df, nzv)$id)) > 0) {
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix="Interact.High.cor.Y",
type=glb_model_type, trainControl.method="repeatedcv",
trainControl.number=glb_rcv_n_folds, trainControl.repeats=glb_rcv_n_repeats,
train.method="glmnet")),
indep_vars=c(max_cor_y_x_vars, paste(max_cor_y_x_vars[1], int_feats, sep=":")),
rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)
} ## [1] "fitting model: Interact.High.cor.Y.glmnet"
## [1] " indep_vars: biddable,startprice,biddable:cellular.fctr"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.55, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = "Interact.High.cor.Y", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 62 -none- numeric
## beta 248 dgCMatrix S4
## df 62 -none- numeric
## dim 2 -none- numeric
## lambda 62 -none- numeric
## dev.ratio 62 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 4 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.098477805 1.095843998 -0.002692227
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.088581422 1.146387537 -0.002860876
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6351931
## 3 0.2 0.6571429
## 4 0.3 0.7105719
## 5 0.4 0.7617021
## 6 0.5 0.7603306
## 7 0.6 0.7500000
## 8 0.7 0.5759494
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Interact.High.cor.Y.glmnet.N
## 1 N 377
## 2 Y 86
## sold.fctr.predict.Interact.High.cor.Y.glmnet.Y
## 1 138
## 2 358
## Prediction
## Reference N Y
## N 377 138
## Y 86 358
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.664234e-01 5.340338e-01 7.383282e-01 7.928718e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 4.865086e-49 6.554149e-04## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6346301
## 3 0.2 0.6517572
## 4 0.3 0.6887052
## 5 0.4 0.7324263
## 6 0.5 0.7522013
## 7 0.6 0.7275168
## 8 0.7 0.5139860
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Interact.High.cor.Y.glmnet.N
## 1 N 404
## 2 Y 117
## sold.fctr.predict.Interact.High.cor.Y.glmnet.Y
## 1 80
## 2 299
## Prediction
## Reference N Y
## N 404 80
## Y 117 299
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.811111e-01 5.569405e-01 7.526434e-01 8.077250e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 4.632073e-52 1.032074e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats
## 1 Interact.High.cor.Y.glmnet biddable,startprice,biddable:cellular.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 25 2.219 0.016
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8567917 0.4 0.7617021 0.7872682
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7383282 0.7928718 0.5695577 0.7664234
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8262948 0.5 0.7522013 0.7811111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7526434 0.807725 0.5569405 0.7811111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01250943 0.0254017# Low.cor.X
# if (glb_is_classification && glb_is_binomial)
# indep_vars_vctr <- subset(glb_feats_df, is.na(cor.high.X) &
# is.ConditionalX.y &
# (exclude.as.feat != 1))[, "id"] else
indep_vars <- subset(glb_feats_df, is.na(cor.high.X) & !nzv &
(exclude.as.feat != 1))[, "id"]
indep_vars <- myadjust_interaction_feats(indep_vars)
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix="Low.cor.X",
type=glb_model_type, trainControl.method="repeatedcv",
trainControl.number=glb_rcv_n_folds, trainControl.repeats=glb_rcv_n_repeats,
train.method="glmnet")),
indep_vars=indep_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)## [1] "fitting model: Low.cor.X.glmnet"
## [1] " indep_vars: biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 1, lambda = 0.00117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = "Low.cor.X", : model's bestTune found at an extreme of
## tuneGrid for parameter: alpha
## Length Class Mode
## a0 83 -none- numeric
## beta 3320 dgCMatrix S4
## df 83 -none- numeric
## dim 2 -none- numeric
## lambda 83 -none- numeric
## dev.ratio 83 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 40 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept)
## 1.180693564
## .rnorm
## 0.009044142
## biddable
## 1.497480151
## cellular.fctr1
## 0.490901786
## cellular.fctrUnknown
## -1.458476711
## color.fctrGold
## -0.313532469
## color.fctrWhite
## -0.029751545
## condition.fctrFor parts or not working
## -0.175408730
## condition.fctrManufacturer refurbished
## -0.191315673
## condition.fctrNew other (see details)
## 0.249073922
## condition.fctrSeller refurbished
## -0.933349976
## prdl.descr.my.fctrUnknown#1
## -0.331779103
## prdl.descr.my.fctriPad1#0
## -0.124076944
## prdl.descr.my.fctriPad1#1
## -0.155117419
## prdl.descr.my.fctriPad2#0
## 1.390608514
## prdl.descr.my.fctriPad2#1
## -0.337242687
## prdl.descr.my.fctriPad3#0
## 0.936090211
## prdl.descr.my.fctriPad3#1
## 0.233944009
## prdl.descr.my.fctriPad4#0
## 1.130305279
## prdl.descr.my.fctriPad4#1
## -1.227643968
## prdl.descr.my.fctriPadAir#0
## 1.775578495
## prdl.descr.my.fctriPadAir#1
## 0.490802449
## prdl.descr.my.fctriPadAir2#0
## 2.627043374
## prdl.descr.my.fctriPadAir2#1
## 2.288130703
## prdl.descr.my.fctriPadmini#0
## 0.005542525
## prdl.descr.my.fctriPadmini#1
## 0.361393096
## prdl.descr.my.fctriPadmini2#0
## 1.387642207
## prdl.descr.my.fctriPadmini2#1
## 0.682004597
## prdl.descr.my.fctriPadmini3#0
## 0.427819092
## prdl.descr.my.fctriPadminiRetina#0
## 2.847067519
## startprice
## -0.010683192
## storage.fctr16
## -0.489462335
## storage.fctr32
## -0.522875231
## storage.fctrUnknown
## 0.740981443
## [1] "max lambda < lambdaOpt:"
## (Intercept)
## 1.19319936
## .rnorm
## 0.01003220
## biddable
## 1.49528038
## cellular.fctr1
## 0.49603851
## cellular.fctrUnknown
## -1.46207317
## color.fctrGold
## -0.32857353
## color.fctrWhite
## -0.03219910
## condition.fctrFor parts or not working
## -0.18210509
## condition.fctrManufacturer refurbished
## -0.19739634
## condition.fctrNew other (see details)
## 0.25395874
## condition.fctrSeller refurbished
## -0.93890819
## prdl.descr.my.fctrUnknown#1
## -0.32134934
## prdl.descr.my.fctriPad1#0
## -0.10611686
## prdl.descr.my.fctriPad1#1
## -0.13676885
## prdl.descr.my.fctriPad2#0
## 1.42438036
## prdl.descr.my.fctriPad2#1
## -0.31504308
## prdl.descr.my.fctriPad3#0
## 0.97158422
## prdl.descr.my.fctriPad3#1
## 0.26523820
## prdl.descr.my.fctriPad4#0
## 1.16509663
## prdl.descr.my.fctriPad4#1
## -1.20418512
## prdl.descr.my.fctriPadAir#0
## 1.81672777
## prdl.descr.my.fctriPadAir#1
## 0.53100360
## prdl.descr.my.fctriPadAir2#0
## 2.68547656
## prdl.descr.my.fctriPadAir2#1
## 2.34603324
## prdl.descr.my.fctriPadmini#0
## 0.03710263
## prdl.descr.my.fctriPadmini#1
## 0.39361956
## prdl.descr.my.fctriPadmini2#0
## 1.42899968
## prdl.descr.my.fctriPadmini2#1
## 0.72459539
## prdl.descr.my.fctriPadmini3#0
## 0.47881757
## prdl.descr.my.fctriPadminiRetina#0
## 2.89424860
## startprice
## -0.01077053
## storage.fctr16
## -0.51705400
## storage.fctr32
## -0.54769432
## storage.fctr64
## -0.02595004
## storage.fctrUnknown
## 0.73616137
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.7072368
## 3 0.2 0.7497725
## 4 0.3 0.7782258
## 5 0.4 0.7938931
## 6 0.5 0.8013937
## 7 0.6 0.7936893
## 8 0.7 0.7704280
## 9 0.8 0.7050562
## 10 0.9 0.4444444
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Low.cor.X.glmnet.N
## 1 N 443
## 2 Y 99
## sold.fctr.predict.Low.cor.X.glmnet.Y
## 1 72
## 2 345
## Prediction
## Reference N Y
## N 443 72
## Y 99 345
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.216893e-01 6.399036e-01 7.959657e-01 8.454193e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 8.991371e-77 4.678187e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6730769
## 3 0.2 0.6981678
## 4 0.3 0.7105538
## 5 0.4 0.7252252
## 6 0.5 0.7436209
## 7 0.6 0.7298335
## 8 0.7 0.6861925
## 9 0.8 0.6234568
## 10 0.9 0.4375000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Low.cor.X.glmnet.N
## 1 N 383
## 2 Y 110
## sold.fctr.predict.Low.cor.X.glmnet.Y
## 1 101
## 2 306
## Prediction
## Reference N Y
## N 383 101
## Y 110 306
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.655556e-01 5.277012e-01 7.364780e-01 7.928829e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.649155e-45 5.818101e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id
## 1 Low.cor.X.glmnet
## feats
## 1 biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 25 2.596 0.083
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8903306 0.5 0.8013937 0.7942268
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7959657 0.8454193 0.5846679 0.8216893
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8131655 0.5 0.7436209 0.7655556
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.736478 0.7928829 0.5277012 0.7655556
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01985552 0.03967914rm(ret_lst)
glb_chunks_df <- myadd_chunk(glb_chunks_df, "fit.models", major.inc=FALSE)## label step_major step_minor label_minor bgn end elapsed
## 10 fit.models 7 0 0 52.896 110.358 57.462
## 11 fit.models 7 1 1 110.358 NA NAfit.models_1_chunk_df <- myadd_chunk(NULL, "fit.models_1_bgn", label.minor="setup")## label step_major step_minor label_minor bgn end elapsed
## 1 fit.models_1_bgn 1 0 setup 119.857 NA NA#stop(here"); glb_to_sav(); glb_models_lst <- sav_models_lst; glb_models_df <- sav_models_df
topindep_var <- NULL; interact_vars <- NULL;
for (mdl_id_pfx in names(glb_mdl_family_lst)) {
fit.models_1_chunk_df <- myadd_chunk(fit.models_1_chunk_df,
paste0("fit.models_1_", mdl_id_pfx), major.inc=TRUE, label.minor="setup")
indep_vars <- NULL;
if (grepl("\\.Interact", mdl_id_pfx)) {
if (is.null(topindep_var) && is.null(interact_vars)) {
# select best glmnet model upto now
dsp_models_df <- orderBy(model_sel_frmla <- get_model_sel_frmla(), glb_models_df)
dsp_models_df <- subset(dsp_models_df, grepl(".glmnet", id, fixed=TRUE))
bst_mdl_id <- dsp_models_df$id[1]
mdl_id_pfx <-
paste(c(head(unlist(strsplit(bst_mdl_id, "[.]")), -1), "Interact"),
collapse=".")
# select most importance feature
if (is.null(bst_featsimp_df <-
myget_feats_importance(glb_models_lst[[bst_mdl_id]]))) {
warning("Base model for RFE.Interact: ", bst_mdl_id,
" has no important features")
next
}
topindep_ix <- 1
while (is.null(topindep_var) && (topindep_ix <= nrow(bst_featsimp_df))) {
topindep_var <- row.names(bst_featsimp_df)[topindep_ix]
if (grepl(".fctr", topindep_var, fixed=TRUE))
topindep_var <- paste0(unlist(strsplit(topindep_var, ".fctr"))[1], ".fctr")
if (topindep_var %in% names(glb_interaction_only_feats_lst)) {
topindep_var <- NULL; topindep_ix <- topindep_ix + 1
} else break
}
# select features with importance > max(10, importance of .rnorm) & is not highest
# combine factor dummy features to just the factor feature
if (length(pos_rnorm <-
grep(".rnorm", row.names(bst_featsimp_df), fixed=TRUE)) > 0)
imp_rnorm <- bst_featsimp_df[pos_rnorm, 1] else
imp_rnorm <- NA
importance_cutoff <- max(10, imp_rnorm, na.rm=TRUE)
interact_vars <-
tail(row.names(subset(bst_featsimp_df, importance > importance_cutoff)), -1)
interact_vars <- myadjust_interaction_feats(myextract_actual_feats(interact_vars))
interact_vars <- interact_vars[!grepl(topindep_var, interact_vars, fixed=TRUE)]
### bid0_sp only
# interact_vars <- c(
# "biddable", "D.ratio.sum.TfIdf.wrds.n", "D.TfIdf.sum.stem.stop.Ratio", "D.sum.TfIdf",
# "D.TfIdf.sum.post.stop", "D.TfIdf.sum.post.stem", "D.ratio.wrds.stop.n.wrds.n", "D.chrs.uppr.n.log",
# "D.chrs.n.log", "color.fctr"
# # , "condition.fctr", "prdl.my.descr.fctr"
# )
# interact_vars <- setdiff(interact_vars, c("startprice.dgt2.is9", "color.fctr"))
###
indep_vars <- myextract_actual_feats(row.names(bst_featsimp_df))
indep_vars <- setdiff(indep_vars, topindep_var)
indep_vars <- setdiff(indep_vars, myextract_actual_feats(interact_vars))
indep_vars <- c(indep_vars,
paste(topindep_var, setdiff(interact_vars, topindep_var), sep="*"))
}
}
if (is.null(indep_vars))
indep_vars <- glb_mdl_feats_lst[[mdl_id_pfx]]
if (is.null(indep_vars) && grepl("RFE\\.", mdl_id_pfx))
indep_vars <- myextract_actual_feats(predictors(rfe_fit_results))
if (is.null(indep_vars))
indep_vars <- subset(glb_feats_df, !nzv & (exclude.as.feat != 1))[, "id"]
indep_vars <- myadjust_interaction_feats(indep_vars)
if (grepl("\\.Interact", mdl_id_pfx)) {
# if (method != tail(unlist(strsplit(bst_mdl_id, "[.]")), 1)) next
if (is.null(glb_mdl_family_lst[[mdl_id_pfx]])) {
if (!is.null(glb_mdl_family_lst[["Best.Interact"]]))
glb_mdl_family_lst[[mdl_id_pfx]] <- glb_mdl_family_lst[["Best.Interact"]]
}
}
if (is.null(glb_mdl_family_lst[[mdl_id_pfx]]))
mdl_methods <- glb_mdl_methods else
mdl_methods <- glb_mdl_family_lst[[mdl_id_pfx]]
for (method in mdl_methods) {
if (method %in% c("rpart", "rf")) {
# rpart: fubar's the tree
# rf: skip the scenario w/ .rnorm for speed
indep_vars <- setdiff(indep_vars, c(".rnorm"))
#mdl_id <- paste0(mdl_id_pfx, ".no.rnorm")
}
fit.models_1_chunk_df <- myadd_chunk(fit.models_1_chunk_df,
paste0("fit.models_1_", mdl_id_pfx), major.inc = FALSE,
label.minor = method)
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix=mdl_id_pfx,
type=glb_model_type, tune.df=glb_tune_models_df,
trainControl.method="repeatedcv",
trainControl.number=glb_rcv_n_folds, trainControl.repeats=glb_rcv_n_repeats,
train.method=method)),
indep_vars=indep_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)
}
}## label step_major step_minor label_minor bgn end
## 1 fit.models_1_bgn 1 0 setup 119.857 119.867
## 2 fit.models_1_RFE.X 2 0 setup 119.868 NA
## elapsed
## 1 0.011
## 2 NA
## label step_major step_minor label_minor bgn end
## 2 fit.models_1_RFE.X 2 0 setup 119.868 119.874
## 3 fit.models_1_RFE.X 2 1 glm 119.875 NA
## elapsed
## 2 0.007
## 3 NA
## [1] "fitting model: RFE.X.glm"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr"
## + Fold1.Rep1: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold1.Rep1: parameter=none
## + Fold2.Rep1: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold2.Rep1: parameter=none
## + Fold3.Rep1: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold3.Rep1: parameter=none
## + Fold1.Rep2: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold1.Rep2: parameter=none
## + Fold2.Rep2: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold2.Rep2: parameter=none
## + Fold3.Rep2: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold3.Rep2: parameter=none
## + Fold1.Rep3: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold1.Rep3: parameter=none
## + Fold2.Rep3: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold2.Rep3: parameter=none
## + Fold3.Rep3: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold3.Rep3: parameter=none
## Aggregating results
## Fitting final model on full training set## Warning: not plotting observations with leverage one:
## 632
## Warning: not plotting observations with leverage one:
## 632
##
## Call:
## NULL
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.8070 -0.6506 -0.1634 0.5519 3.6522
##
## Coefficients: (15 not defined because of singularities)
## Estimate Std. Error z value
## (Intercept) 1.527309 0.842013 1.814
## .rnorm 0.048634 0.090837 0.535
## biddable 1.466504 0.203525 7.206
## cellular.fctr1 0.099985 0.293406 0.341
## cellular.fctrUnknown -1.405923 0.529180 -2.657
## color.fctrGold -0.545864 0.773849 -0.705
## `color.fctrSpace Gray` -0.079646 0.367823 -0.217
## color.fctrUnknown -0.035726 0.262761 -0.136
## color.fctrWhite -0.075860 0.296990 -0.255
## `condition.fctrFor parts or not working` -0.323789 0.322695 -1.003
## `condition.fctrManufacturer refurbished` -0.373566 0.707124 -0.528
## condition.fctrNew -0.036170 0.345302 -0.105
## `condition.fctrNew other (see details)` 0.232572 0.502264 0.463
## `condition.fctrSeller refurbished` -0.970790 0.410514 -2.365
## `prdl.descr.my.fctrUnknown#1` -0.081642 0.597144 -0.137
## `prdl.descr.my.fctriPad1#0` 0.148480 0.544708 0.273
## `prdl.descr.my.fctriPad1#1` 0.177083 0.555149 0.319
## `prdl.descr.my.fctriPad2#0` 1.853308 0.675704 2.743
## `prdl.descr.my.fctriPad2#1` -0.077611 0.546213 -0.142
## `prdl.descr.my.fctriPad3#0` 1.341347 0.680622 1.971
## `prdl.descr.my.fctriPad3#1` 0.706333 0.572402 1.234
## `prdl.descr.my.fctriPad4#0` 1.673896 0.625625 2.676
## `prdl.descr.my.fctriPad4#1` -0.966609 0.841416 -1.149
## `prdl.descr.my.fctriPad5#0` NA NA NA
## `prdl.descr.my.fctriPadAir#0` 2.363317 0.670859 3.523
## `prdl.descr.my.fctriPadAir#1` 1.001925 0.670291 1.495
## `prdl.descr.my.fctriPadAir2#0` 3.455228 0.741144 4.662
## `prdl.descr.my.fctriPadAir2#1` 3.202026 0.870236 3.679
## `prdl.descr.my.fctriPadmini#0` 0.458368 0.517834 0.885
## `prdl.descr.my.fctriPadmini#1` 0.824096 0.582674 1.414
## `prdl.descr.my.fctriPadmini2#0` 1.876058 0.674030 2.783
## `prdl.descr.my.fctriPadmini2#1` 0.967575 0.898406 1.077
## `prdl.descr.my.fctriPadmini3#0` 1.170204 0.776048 1.508
## `prdl.descr.my.fctriPadmini3#1` -10.119782 614.033655 -0.016
## `prdl.descr.my.fctriPadminiRetina#0` 3.541400 1.222350 2.897
## `prdl.descr.my.fctriPadminiRetina#1` NA NA NA
## startprice -0.012203 0.001341 -9.101
## storage.fctr16 -0.952616 0.634389 -1.502
## storage.fctr32 -0.949899 0.648002 -1.466
## storage.fctr64 -0.427307 0.643610 -0.664
## storage.fctrUnknown 0.432248 0.791122 0.546
## `cellular.fctr0:carrier.fctrNone` NA NA NA
## `cellular.fctr1:carrier.fctrNone` NA NA NA
## `cellular.fctrUnknown:carrier.fctrNone` NA NA NA
## `cellular.fctr0:carrier.fctrOther` NA NA NA
## `cellular.fctr1:carrier.fctrOther` 12.196891 882.743862 0.014
## `cellular.fctrUnknown:carrier.fctrOther` NA NA NA
## `cellular.fctr0:carrier.fctrSprint` NA NA NA
## `cellular.fctr1:carrier.fctrSprint` 1.046346 0.943176 1.109
## `cellular.fctrUnknown:carrier.fctrSprint` NA NA NA
## `cellular.fctr0:carrier.fctrT-Mobile` NA NA NA
## `cellular.fctr1:carrier.fctrT-Mobile` 1.849648 0.883572 2.093
## `cellular.fctrUnknown:carrier.fctrT-Mobile` NA NA NA
## `cellular.fctr0:carrier.fctrUnknown` NA NA NA
## `cellular.fctr1:carrier.fctrUnknown` 0.881239 0.494560 1.782
## `cellular.fctrUnknown:carrier.fctrUnknown` NA NA NA
## `cellular.fctr0:carrier.fctrVerizon` NA NA NA
## `cellular.fctr1:carrier.fctrVerizon` 0.678137 0.451610 1.502
## `cellular.fctrUnknown:carrier.fctrVerizon` NA NA NA
## Pr(>|z|)
## (Intercept) 0.069696 .
## .rnorm 0.592377
## biddable 5.78e-13 ***
## cellular.fctr1 0.733273
## cellular.fctrUnknown 0.007889 **
## color.fctrGold 0.480568
## `color.fctrSpace Gray` 0.828572
## color.fctrUnknown 0.891850
## color.fctrWhite 0.798392
## `condition.fctrFor parts or not working` 0.315672
## `condition.fctrManufacturer refurbished` 0.597298
## condition.fctrNew 0.916575
## `condition.fctrNew other (see details)` 0.643331
## `condition.fctrSeller refurbished` 0.018039 *
## `prdl.descr.my.fctrUnknown#1` 0.891252
## `prdl.descr.my.fctriPad1#0` 0.785172
## `prdl.descr.my.fctriPad1#1` 0.749740
## `prdl.descr.my.fctriPad2#0` 0.006092 **
## `prdl.descr.my.fctriPad2#1` 0.887009
## `prdl.descr.my.fctriPad3#0` 0.048751 *
## `prdl.descr.my.fctriPad3#1` 0.217210
## `prdl.descr.my.fctriPad4#0` 0.007461 **
## `prdl.descr.my.fctriPad4#1` 0.250643
## `prdl.descr.my.fctriPad5#0` NA
## `prdl.descr.my.fctriPadAir#0` 0.000427 ***
## `prdl.descr.my.fctriPadAir#1` 0.134977
## `prdl.descr.my.fctriPadAir2#0` 3.13e-06 ***
## `prdl.descr.my.fctriPadAir2#1` 0.000234 ***
## `prdl.descr.my.fctriPadmini#0` 0.376068
## `prdl.descr.my.fctriPadmini#1` 0.157264
## `prdl.descr.my.fctriPadmini2#0` 0.005380 **
## `prdl.descr.my.fctriPadmini2#1` 0.281484
## `prdl.descr.my.fctriPadmini3#0` 0.131580
## `prdl.descr.my.fctriPadmini3#1` 0.986851
## `prdl.descr.my.fctriPadminiRetina#0` 0.003765 **
## `prdl.descr.my.fctriPadminiRetina#1` NA
## startprice < 2e-16 ***
## storage.fctr16 0.133193
## storage.fctr32 0.142678
## storage.fctr64 0.506740
## storage.fctrUnknown 0.584810
## `cellular.fctr0:carrier.fctrNone` NA
## `cellular.fctr1:carrier.fctrNone` NA
## `cellular.fctrUnknown:carrier.fctrNone` NA
## `cellular.fctr0:carrier.fctrOther` NA
## `cellular.fctr1:carrier.fctrOther` 0.988976
## `cellular.fctrUnknown:carrier.fctrOther` NA
## `cellular.fctr0:carrier.fctrSprint` NA
## `cellular.fctr1:carrier.fctrSprint` 0.267264
## `cellular.fctrUnknown:carrier.fctrSprint` NA
## `cellular.fctr0:carrier.fctrT-Mobile` NA
## `cellular.fctr1:carrier.fctrT-Mobile` 0.036316 *
## `cellular.fctrUnknown:carrier.fctrT-Mobile` NA
## `cellular.fctr0:carrier.fctrUnknown` NA
## `cellular.fctr1:carrier.fctrUnknown` 0.074771 .
## `cellular.fctrUnknown:carrier.fctrUnknown` NA
## `cellular.fctr0:carrier.fctrVerizon` NA
## `cellular.fctr1:carrier.fctrVerizon` 0.133200
## `cellular.fctrUnknown:carrier.fctrVerizon` NA
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1324.19 on 958 degrees of freedom
## Residual deviance: 783.87 on 915 degrees of freedom
## AIC: 871.87
##
## Number of Fisher Scoring iterations: 13
##
## [1] " calling mypredict_mdl for fit:"## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.7164179
## 3 0.2 0.7550645
## 4 0.3 0.7746479
## 5 0.4 0.8000000
## 6 0.5 0.7981330
## 7 0.6 0.8044010
## 8 0.7 0.7708333
## 9 0.8 0.7088608
## 10 0.9 0.4890388
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.6000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.RFE.X.glm.N sold.fctr.predict.RFE.X.glm.Y
## 1 N 470 45
## 2 Y 115 329
## Prediction
## Reference N Y
## N 470 45
## Y 115 329
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.331595e-01 6.607936e-01 8.080269e-01 8.562261e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 1.491981e-83 4.899243e-08## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only## [1] " calling mypredict_mdl for OOB:"## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6761313
## 3 0.2 0.6887805
## 4 0.3 0.7104984
## 5 0.4 0.7280899
## 6 0.5 0.7406514
## 7 0.6 0.7321429
## 8 0.7 0.6851595
## 9 0.8 0.6253776
## 10 0.9 0.4428312
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.RFE.X.glm.N sold.fctr.predict.RFE.X.glm.Y
## 1 N 378 106
## 2 Y 109 307
## Prediction
## Reference N Y
## N 378 106
## Y 109 307
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.611111e-01 5.192355e-01 7.318690e-01 7.886326e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 9.811484e-44 8.915060e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only## id
## 1 RFE.X.glm
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 1 1.844 0.087
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8930683 0.6 0.804401 0.7900546
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.8080269 0.8562261 0.5764108 0.8331595
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8055666 0.5 0.7406514 0.7611111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.731869 0.7886326 0.5192355 0.7611111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01572842 0.03141904
## label step_major step_minor label_minor bgn end
## 3 fit.models_1_RFE.X 2 1 glm 119.875 125.572
## 4 fit.models_1_RFE.X 2 2 bayesglm 125.573 NA
## elapsed
## 3 5.697
## 4 NA
## [1] "fitting model: RFE.X.bayesglm"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr"## Loading required package: arm
## Loading required package: MASS
##
## Attaching package: 'MASS'
##
## The following object is masked from 'package:dplyr':
##
## select
##
## Loading required package: lme4
##
## arm (Version 1.8-6, built: 2015-7-7)
##
## Working directory is /Users/bbalaji-2012/Documents/Work/Courses/MIT/Analytics_Edge_15_071x/Assignments/Kaggle_eBay_iPads
## + Fold1.Rep1: parameter=none
## - Fold1.Rep1: parameter=none
## + Fold2.Rep1: parameter=none
## - Fold2.Rep1: parameter=none
## + Fold3.Rep1: parameter=none
## - Fold3.Rep1: parameter=none
## + Fold1.Rep2: parameter=none
## - Fold1.Rep2: parameter=none
## + Fold2.Rep2: parameter=none
## - Fold2.Rep2: parameter=none
## + Fold3.Rep2: parameter=none
## - Fold3.Rep2: parameter=none
## + Fold1.Rep3: parameter=none
## - Fold1.Rep3: parameter=none
## + Fold2.Rep3: parameter=none
## - Fold2.Rep3: parameter=none
## + Fold3.Rep3: parameter=none
## - Fold3.Rep3: parameter=none
## Aggregating results
## Fitting final model on full training set
##
## Call:
## NULL
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.7000 -0.6625 -0.1939 0.5765 3.4697
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 1.381226 1.462489 0.944
## .rnorm 0.039732 0.088785 0.448
## biddable 1.492119 0.198137 7.531
## cellular.fctr1 0.349056 1.299723 0.269
## cellular.fctrUnknown -0.670616 1.686255 -0.398
## color.fctrGold -0.367656 0.677668 -0.543
## `color.fctrSpace Gray` -0.030909 0.350778 -0.088
## color.fctrUnknown -0.003907 0.251943 -0.016
## color.fctrWhite -0.045438 0.283871 -0.160
## `condition.fctrFor parts or not working` -0.258200 0.311399 -0.829
## `condition.fctrManufacturer refurbished` -0.330185 0.654715 -0.504
## condition.fctrNew -0.045651 0.330699 -0.138
## `condition.fctrNew other (see details)` 0.218212 0.476067 0.458
## `condition.fctrSeller refurbished` -0.905440 0.395683 -2.288
## `prdl.descr.my.fctrUnknown#1` -0.361189 0.526891 -0.686
## `prdl.descr.my.fctriPad1#0` -0.236775 0.448236 -0.528
## `prdl.descr.my.fctriPad1#1` -0.221093 0.459072 -0.482
## `prdl.descr.my.fctriPad2#0` 1.305779 0.569833 2.292
## `prdl.descr.my.fctriPad2#1` -0.494174 0.442130 -1.118
## `prdl.descr.my.fctriPad3#0` 0.797209 0.566697 1.407
## `prdl.descr.my.fctriPad3#1` 0.212243 0.465850 0.456
## `prdl.descr.my.fctriPad4#0` 1.092840 0.515833 2.119
## `prdl.descr.my.fctriPad4#1` -1.345568 0.694620 -1.937
## `prdl.descr.my.fctriPad5#0` 0.000000 2.500000 0.000
## `prdl.descr.my.fctriPadAir#0` 1.693095 0.544699 3.108
## `prdl.descr.my.fctriPadAir#1` 0.402444 0.542095 0.742
## `prdl.descr.my.fctriPadAir2#0` 2.608403 0.597371 4.366
## `prdl.descr.my.fctriPadAir2#1` 2.305341 0.714819 3.225
## `prdl.descr.my.fctriPadmini#0` -0.035991 0.406448 -0.089
## `prdl.descr.my.fctriPadmini#1` 0.313264 0.471860 0.664
## `prdl.descr.my.fctriPadmini2#0` 1.236774 0.556311 2.223
## `prdl.descr.my.fctriPadmini2#1` 0.380766 0.754266 0.505
## `prdl.descr.my.fctriPadmini3#0` 0.473279 0.640112 0.739
## `prdl.descr.my.fctriPadmini3#1` -0.235435 2.088252 -0.113
## `prdl.descr.my.fctriPadminiRetina#0` 2.484276 1.078128 2.304
## `prdl.descr.my.fctriPadminiRetina#1` 0.000000 2.500000 0.000
## startprice -0.011128 0.001191 -9.344
## storage.fctr16 -0.729506 0.528230 -1.381
## storage.fctr32 -0.751807 0.543453 -1.383
## storage.fctr64 -0.236144 0.541853 -0.436
## storage.fctrUnknown 0.522825 0.670939 0.779
## `cellular.fctr0:carrier.fctrNone` 0.215070 1.295365 0.166
## `cellular.fctr1:carrier.fctrNone` 0.000000 2.500000 0.000
## `cellular.fctrUnknown:carrier.fctrNone` 0.000000 2.500000 0.000
## `cellular.fctr0:carrier.fctrOther` 0.000000 2.500000 0.000
## `cellular.fctr1:carrier.fctrOther` 0.362403 1.971100 0.184
## `cellular.fctrUnknown:carrier.fctrOther` 0.000000 2.500000 0.000
## `cellular.fctr0:carrier.fctrSprint` 0.000000 2.500000 0.000
## `cellular.fctr1:carrier.fctrSprint` 0.831634 0.838362 0.992
## `cellular.fctrUnknown:carrier.fctrSprint` 0.000000 2.500000 0.000
## `cellular.fctr0:carrier.fctrT-Mobile` 0.000000 2.500000 0.000
## `cellular.fctr1:carrier.fctrT-Mobile` 1.510817 0.807764 1.870
## `cellular.fctrUnknown:carrier.fctrT-Mobile` 0.000000 2.500000 0.000
## `cellular.fctr0:carrier.fctrUnknown` 0.000000 2.500000 0.000
## `cellular.fctr1:carrier.fctrUnknown` 0.739794 0.465708 1.589
## `cellular.fctrUnknown:carrier.fctrUnknown` -0.670616 1.686255 -0.398
## `cellular.fctr0:carrier.fctrVerizon` 0.000000 2.500000 0.000
## `cellular.fctr1:carrier.fctrVerizon` 0.586465 0.425240 1.379
## `cellular.fctrUnknown:carrier.fctrVerizon` 0.000000 2.500000 0.000
## Pr(>|z|)
## (Intercept) 0.34495
## .rnorm 0.65451
## biddable 5.05e-14 ***
## cellular.fctr1 0.78827
## cellular.fctrUnknown 0.69085
## color.fctrGold 0.58745
## `color.fctrSpace Gray` 0.92978
## color.fctrUnknown 0.98763
## color.fctrWhite 0.87283
## `condition.fctrFor parts or not working` 0.40701
## `condition.fctrManufacturer refurbished` 0.61404
## condition.fctrNew 0.89021
## `condition.fctrNew other (see details)` 0.64669
## `condition.fctrSeller refurbished` 0.02212 *
## `prdl.descr.my.fctrUnknown#1` 0.49302
## `prdl.descr.my.fctriPad1#0` 0.59733
## `prdl.descr.my.fctriPad1#1` 0.63008
## `prdl.descr.my.fctriPad2#0` 0.02193 *
## `prdl.descr.my.fctriPad2#1` 0.26369
## `prdl.descr.my.fctriPad3#0` 0.15950
## `prdl.descr.my.fctriPad3#1` 0.64867
## `prdl.descr.my.fctriPad4#0` 0.03412 *
## `prdl.descr.my.fctriPad4#1` 0.05273 .
## `prdl.descr.my.fctriPad5#0` 1.00000
## `prdl.descr.my.fctriPadAir#0` 0.00188 **
## `prdl.descr.my.fctriPadAir#1` 0.45785
## `prdl.descr.my.fctriPadAir2#0` 1.26e-05 ***
## `prdl.descr.my.fctriPadAir2#1` 0.00126 **
## `prdl.descr.my.fctriPadmini#0` 0.92944
## `prdl.descr.my.fctriPadmini#1` 0.50676
## `prdl.descr.my.fctriPadmini2#0` 0.02620 *
## `prdl.descr.my.fctriPadmini2#1` 0.61369
## `prdl.descr.my.fctriPadmini3#0` 0.45968
## `prdl.descr.my.fctriPadmini3#1` 0.91023
## `prdl.descr.my.fctriPadminiRetina#0` 0.02121 *
## `prdl.descr.my.fctriPadminiRetina#1` 1.00000
## startprice < 2e-16 ***
## storage.fctr16 0.16727
## storage.fctr32 0.16655
## storage.fctr64 0.66298
## storage.fctrUnknown 0.43584
## `cellular.fctr0:carrier.fctrNone` 0.86813
## `cellular.fctr1:carrier.fctrNone` 1.00000
## `cellular.fctrUnknown:carrier.fctrNone` 1.00000
## `cellular.fctr0:carrier.fctrOther` 1.00000
## `cellular.fctr1:carrier.fctrOther` 0.85412
## `cellular.fctrUnknown:carrier.fctrOther` 1.00000
## `cellular.fctr0:carrier.fctrSprint` 1.00000
## `cellular.fctr1:carrier.fctrSprint` 0.32121
## `cellular.fctrUnknown:carrier.fctrSprint` 1.00000
## `cellular.fctr0:carrier.fctrT-Mobile` 1.00000
## `cellular.fctr1:carrier.fctrT-Mobile` 0.06143 .
## `cellular.fctrUnknown:carrier.fctrT-Mobile` 1.00000
## `cellular.fctr0:carrier.fctrUnknown` 1.00000
## `cellular.fctr1:carrier.fctrUnknown` 0.11217
## `cellular.fctrUnknown:carrier.fctrUnknown` 0.69085
## `cellular.fctr0:carrier.fctrVerizon` 1.00000
## `cellular.fctr1:carrier.fctrVerizon` 0.16785
## `cellular.fctrUnknown:carrier.fctrVerizon` 1.00000
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1324.19 on 958 degrees of freedom
## Residual deviance: 786.44 on 900 degrees of freedom
## AIC: 904.44
##
## Number of Fisher Scoring iterations: 11
##
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.7120591
## 3 0.2 0.7557182
## 4 0.3 0.7751004
## 5 0.4 0.7934426
## 6 0.5 0.7972028
## 7 0.6 0.7960928
## 8 0.7 0.7688312
## 9 0.8 0.7032349
## 10 0.9 0.4551724
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.RFE.X.bayesglm.N
## 1 N 443
## 2 Y 102
## sold.fctr.predict.RFE.X.bayesglm.Y
## 1 72
## 2 342
## Prediction
## Reference N Y
## N 443 72
## Y 102 342
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.185610e-01 6.334146e-01 7.926828e-01 8.424654e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 5.463537e-75 2.791461e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6736475
## 3 0.2 0.6937198
## 4 0.3 0.7094737
## 5 0.4 0.7266592
## 6 0.5 0.7442424
## 7 0.6 0.7330779
## 8 0.7 0.6934813
## 9 0.8 0.6196319
## 10 0.9 0.4309392
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.RFE.X.bayesglm.N
## 1 N 382
## 2 Y 109
## sold.fctr.predict.RFE.X.bayesglm.Y
## 1 102
## 2 307
## Prediction
## Reference N Y
## N 382 102
## Y 109 307
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.655556e-01 5.278609e-01 7.364780e-01 7.928829e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.649155e-45 6.795648e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id
## 1 RFE.X.bayesglm
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 1 2.819 0.125
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8919881 0.5 0.7972028 0.790404
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7926828 0.8424654 0.5769613 0.818561
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8111689 0.5 0.7442424 0.7655556
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.736478 0.7928829 0.5278609 0.7655556
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01825537 0.03653979
## label step_major step_minor label_minor bgn end
## 4 fit.models_1_RFE.X 2 2 bayesglm 125.573 131.602
## 5 fit.models_1_RFE.X 2 3 glmnet 131.602 NA
## elapsed
## 4 6.029
## 5 NA
## [1] "fitting model: RFE.X.glmnet"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.55, lambda = 0.0252 on full training set
## Length Class Mode
## a0 86 -none- numeric
## beta 4988 dgCMatrix S4
## df 86 -none- numeric
## dim 2 -none- numeric
## lambda 86 -none- numeric
## dev.ratio 86 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 58 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept)
## 0.212280420
## biddable
## 1.517695590
## cellular.fctr1
## 0.111291692
## cellular.fctrUnknown
## -0.240821550
## condition.fctrSeller refurbished
## -0.303613584
## prdl.descr.my.fctrUnknown#1
## -0.001428038
## prdl.descr.my.fctriPad2#0
## 0.578460972
## prdl.descr.my.fctriPad2#1
## -0.138431499
## prdl.descr.my.fctriPad3#0
## 0.193090727
## prdl.descr.my.fctriPad4#0
## 0.239392597
## prdl.descr.my.fctriPad4#1
## -0.926942426
## prdl.descr.my.fctriPadAir#0
## 0.356239530
## prdl.descr.my.fctriPadAir2#0
## 0.496926272
## prdl.descr.my.fctriPadAir2#1
## 0.329623264
## prdl.descr.my.fctriPadmini#0
## -0.054182251
## prdl.descr.my.fctriPadmini2#0
## 0.071308286
## prdl.descr.my.fctriPadmini3#0
## -0.058991667
## prdl.descr.my.fctriPadminiRetina#0
## 0.615538025
## startprice
## -0.005378116
## storage.fctr32
## -0.012610177
## storage.fctr64
## 0.062650081
## cellular.fctr1:carrier.fctrT-Mobile
## 0.198632683
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.232792402
## cellular.fctr1:carrier.fctrVerizon
## 0.039311991
## [1] "max lambda < lambdaOpt:"
## (Intercept)
## 0.242008407
## biddable
## 1.519615706
## cellular.fctr1
## 0.118581035
## cellular.fctrUnknown
## -0.251456870
## condition.fctrSeller refurbished
## -0.341299917
## prdl.descr.my.fctrUnknown#1
## -0.016706048
## prdl.descr.my.fctriPad2#0
## 0.620952079
## prdl.descr.my.fctriPad2#1
## -0.167333182
## prdl.descr.my.fctriPad3#0
## 0.225106750
## prdl.descr.my.fctriPad4#0
## 0.288944274
## prdl.descr.my.fctriPad4#1
## -0.963264406
## prdl.descr.my.fctriPadAir#0
## 0.426395470
## prdl.descr.my.fctriPadAir2#0
## 0.587153313
## prdl.descr.my.fctriPadAir2#1
## 0.421222039
## prdl.descr.my.fctriPadmini#0
## -0.066928131
## prdl.descr.my.fctriPadmini2#0
## 0.133141513
## prdl.descr.my.fctriPadmini3#0
## -0.059558089
## prdl.descr.my.fctriPadminiRetina#0
## 0.775062295
## startprice
## -0.005612763
## storage.fctr32
## -0.029076706
## storage.fctr64
## 0.078190707
## cellular.fctr1:carrier.fctrT-Mobile
## 0.294925866
## cellular.fctr1:carrier.fctrUnknown
## 0.024140616
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.241582561
## cellular.fctr1:carrier.fctrVerizon
## 0.069556614
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6666667
## 3 0.2 0.7179487
## 4 0.3 0.7579556
## 5 0.4 0.7662338
## 6 0.5 0.7792208
## 7 0.6 0.7660099
## 8 0.7 0.7591623
## 9 0.8 0.5931677
## 10 0.9 0.0944206
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.RFE.X.glmnet.N
## 1 N 442
## 2 Y 114
## sold.fctr.predict.RFE.X.glmnet.Y
## 1 73
## 2 330
## Prediction
## Reference N Y
## N 442 73
## Y 114 330
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.050052e-01 6.053489e-01 7.784873e-01 8.296347e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 1.430019e-67 3.443570e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.63221884
## 2 0.1 0.65329053
## 3 0.2 0.68177697
## 4 0.3 0.70647773
## 5 0.4 0.73198198
## 6 0.5 0.75735294
## 7 0.6 0.73850197
## 8 0.7 0.69132290
## 9 0.8 0.51027397
## 10 0.9 0.08715596
## 11 1.0 0.00000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.RFE.X.glmnet.N
## 1 N 393
## 2 Y 107
## sold.fctr.predict.RFE.X.glmnet.Y
## 1 91
## 2 309
## Prediction
## Reference N Y
## N 393 91
## Y 107 309
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.800000e-01 5.562749e-01 7.514869e-01 8.066667e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.417978e-51 2.864220e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id
## 1 RFE.X.glmnet
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 25 2.675 0.096
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8767777 0.5 0.7792208 0.7949168
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7784873 0.8296347 0.5850163 0.8050052
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8178863 0.5 0.7573529 0.78
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7514869 0.8066667 0.5562749 0.78
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.008376313 0.01748681
## label step_major step_minor label_minor bgn end
## 5 fit.models_1_RFE.X 2 3 glmnet 131.602 138.244
## 6 fit.models_1_RFE.X 2 4 rpart 138.244 NA
## elapsed
## 5 6.642
## 6 NA
## [1] "fitting model: RFE.X.rpart"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr"
## Aggregating results
## Selecting tuning parameters
## Fitting cp = 0.00901 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = mdl_id_pfx, : model's bestTune found at an extreme of
## tuneGrid for parameter: cp
## Call:
## rpart(formula = .outcome ~ ., control = list(minsplit = 20, minbucket = 7,
## cp = 0, maxcompete = 4, maxsurrogate = 5, usesurrogate = 2,
## surrogatestyle = 0, maxdepth = 30, xval = 0))
## n= 959
##
## CP nsplit rel error
## 1 0.511261261 0 1.0000000
## 2 0.027027027 1 0.4887387
## 3 0.015765766 3 0.4346847
## 4 0.011261261 4 0.4189189
## 5 0.009009009 5 0.4076577
##
## Variable importance
## biddable
## 46
## startprice
## 35
## prdl.descr.my.fctriPad3#0
## 3
## condition.fctrFor parts or not working
## 3
## prdl.descr.my.fctriPad2#0
## 2
## condition.fctrNew
## 2
## prdl.descr.my.fctriPadAir#0
## 2
## color.fctrUnknown
## 1
## prdl.descr.my.fctriPadAir2#0
## 1
## prdl.descr.my.fctriPad1#0
## 1
## prdl.descr.my.fctriPadmini3#0
## 1
## prdl.descr.my.fctriPadAir#1
## 1
##
## Node number 1: 959 observations, complexity param=0.5112613
## predicted class=N expected loss=0.4629823 P(node) =1
## class counts: 515 444
## probabilities: 0.537 0.463
## left son=2 (526 obs) right son=3 (433 obs)
## Primary splits:
## biddable < 0.5 to the left, improve=141.28850, (0 missing)
## startprice < 100.5 to the right, improve=136.23370, (0 missing)
## condition.fctrNew < 0.5 to the right, improve= 15.58147, (0 missing)
## prdl.descr.my.fctriPad2#0 < 0.5 to the left, improve= 13.17333, (0 missing)
## prdl.descr.my.fctriPad4#1 < 0.5 to the right, improve= 10.99927, (0 missing)
## Surrogate splits:
## startprice < 100.5 to the right, agree=0.755, adj=0.457, (0 split)
## prdl.descr.my.fctriPad3#0 < 0.5 to the left, agree=0.579, adj=0.067, (0 split)
## condition.fctrFor parts or not working < 0.5 to the left, agree=0.572, adj=0.053, (0 split)
## prdl.descr.my.fctriPad2#0 < 0.5 to the left, agree=0.571, adj=0.051, (0 split)
## prdl.descr.my.fctriPad1#0 < 0.5 to the left, agree=0.557, adj=0.018, (0 split)
##
## Node number 2: 526 observations
## predicted class=N expected loss=0.21673 P(node) =0.548488
## class counts: 412 114
## probabilities: 0.783 0.217
##
## Node number 3: 433 observations, complexity param=0.02702703
## predicted class=Y expected loss=0.2378753 P(node) =0.451512
## class counts: 103 330
## probabilities: 0.238 0.762
## left son=6 (153 obs) right son=7 (280 obs)
## Primary splits:
## startprice < 129.995 to the right, improve=38.436390, (0 missing)
## prdl.descr.my.fctriPad4#1 < 0.5 to the right, improve= 6.469085, (0 missing)
## condition.fctrNew < 0.5 to the right, improve= 4.644165, (0 missing)
## prdl.descr.my.fctriPadmini3#0 < 0.5 to the right, improve= 3.501587, (0 missing)
## prdl.descr.my.fctriPad2#0 < 0.5 to the left, improve= 3.078297, (0 missing)
## Surrogate splits:
## condition.fctrNew < 0.5 to the right, agree=0.709, adj=0.176, (0 split)
## prdl.descr.my.fctriPadAir2#0 < 0.5 to the right, agree=0.674, adj=0.078, (0 split)
## prdl.descr.my.fctriPadmini3#0 < 0.5 to the right, agree=0.670, adj=0.065, (0 split)
## prdl.descr.my.fctriPadAir#0 < 0.5 to the right, agree=0.661, adj=0.039, (0 split)
## prdl.descr.my.fctriPadAir#1 < 0.5 to the right, agree=0.661, adj=0.039, (0 split)
##
## Node number 6: 153 observations, complexity param=0.02702703
## predicted class=N expected loss=0.4771242 P(node) =0.1595412
## class counts: 80 73
## probabilities: 0.523 0.477
## left son=12 (92 obs) right son=13 (61 obs)
## Primary splits:
## startprice < 205 to the right, improve=5.339157, (0 missing)
## prdl.descr.my.fctriPad4#1 < 0.5 to the right, improve=3.339869, (0 missing)
## cellular.fctr1 < 0.5 to the left, improve=3.032296, (0 missing)
## prdl.descr.my.fctriPadAir#0 < 0.5 to the left, improve=2.934936, (0 missing)
## cellular.fctr1:carrier.fctrVerizon < 0.5 to the left, improve=2.757283, (0 missing)
## Surrogate splits:
## storage.fctr16 < 0.5 to the left, agree=0.660, adj=0.148, (0 split)
## prdl.descr.my.fctriPad3#0 < 0.5 to the left, agree=0.654, adj=0.131, (0 split)
## prdl.descr.my.fctriPad2#0 < 0.5 to the left, agree=0.627, adj=0.066, (0 split)
## prdl.descr.my.fctriPadAir#1 < 0.5 to the left, agree=0.627, adj=0.066, (0 split)
## prdl.descr.my.fctriPad4#0 < 0.5 to the left, agree=0.621, adj=0.049, (0 split)
##
## Node number 7: 280 observations
## predicted class=Y expected loss=0.08214286 P(node) =0.2919708
## class counts: 23 257
## probabilities: 0.082 0.918
##
## Node number 12: 92 observations, complexity param=0.01576577
## predicted class=N expected loss=0.3695652 P(node) =0.09593326
## class counts: 58 34
## probabilities: 0.630 0.370
## left son=24 (81 obs) right son=25 (11 obs)
## Primary splits:
## prdl.descr.my.fctriPadAir#0 < 0.5 to the left, improve=5.028937, (0 missing)
## startprice < 240.995 to the left, improve=2.646820, (0 missing)
## prdl.descr.my.fctriPadmini#0 < 0.5 to the right, improve=1.332751, (0 missing)
## storage.fctr16 < 0.5 to the right, improve=1.185995, (0 missing)
## storage.fctr32 < 0.5 to the left, improve=1.176703, (0 missing)
##
## Node number 13: 61 observations, complexity param=0.01126126
## predicted class=Y expected loss=0.3606557 P(node) =0.06360792
## class counts: 22 39
## probabilities: 0.361 0.639
## left son=26 (23 obs) right son=27 (38 obs)
## Primary splits:
## color.fctrUnknown < 0.5 to the right, improve=4.543047, (0 missing)
## cellular.fctr1 < 0.5 to the left, improve=4.512100, (0 missing)
## storage.fctr16 < 0.5 to the left, improve=3.203499, (0 missing)
## color.fctrWhite < 0.5 to the left, improve=3.180243, (0 missing)
## startprice < 157.495 to the left, improve=1.414372, (0 missing)
## Surrogate splits:
## condition.fctrFor parts or not working < 0.5 to the right, agree=0.689, adj=0.174, (0 split)
## prdl.descr.my.fctriPadmini#1 < 0.5 to the right, agree=0.689, adj=0.174, (0 split)
## storage.fctrUnknown < 0.5 to the right, agree=0.689, adj=0.174, (0 split)
## color.fctrWhite < 0.5 to the left, agree=0.672, adj=0.130, (0 split)
## cellular.fctrUnknown < 0.5 to the right, agree=0.656, adj=0.087, (0 split)
##
## Node number 24: 81 observations
## predicted class=N expected loss=0.308642 P(node) =0.08446298
## class counts: 56 25
## probabilities: 0.691 0.309
##
## Node number 25: 11 observations
## predicted class=Y expected loss=0.1818182 P(node) =0.01147028
## class counts: 2 9
## probabilities: 0.182 0.818
##
## Node number 26: 23 observations
## predicted class=N expected loss=0.3913043 P(node) =0.02398332
## class counts: 14 9
## probabilities: 0.609 0.391
##
## Node number 27: 38 observations
## predicted class=Y expected loss=0.2105263 P(node) =0.03962461
## class counts: 8 30
## probabilities: 0.211 0.789
##
## n= 959
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 959 444 N (0.53701773 0.46298227)
## 2) biddable< 0.5 526 114 N (0.78326996 0.21673004) *
## 3) biddable>=0.5 433 103 Y (0.23787529 0.76212471)
## 6) startprice>=129.995 153 73 N (0.52287582 0.47712418)
## 12) startprice>=205 92 34 N (0.63043478 0.36956522)
## 24) prdl.descr.my.fctriPadAir#0< 0.5 81 25 N (0.69135802 0.30864198) *
## 25) prdl.descr.my.fctriPadAir#0>=0.5 11 2 Y (0.18181818 0.81818182) *
## 13) startprice< 205 61 22 Y (0.36065574 0.63934426)
## 26) color.fctrUnknown>=0.5 23 9 N (0.60869565 0.39130435) *
## 27) color.fctrUnknown< 0.5 38 8 Y (0.21052632 0.78947368) *
## 7) startprice< 129.995 280 23 Y (0.08214286 0.91785714) *
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6329294
## 3 0.2 0.6329294
## 4 0.3 0.7525656
## 5 0.4 0.7658473
## 6 0.5 0.7658473
## 7 0.6 0.7658473
## 8 0.7 0.7658473
## 9 0.8 0.7238095
## 10 0.9 0.7099448
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.7000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.RFE.X.rpart.N
## 1 N 482
## 2 Y 148
## sold.fctr.predict.RFE.X.rpart.Y
## 1 33
## 2 296
## Prediction
## Reference N Y
## N 482 33
## Y 148 296
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.112617e-01 6.135432e-01 7.850331e-01 8.355626e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 6.218110e-71 2.380189e-17## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6322188
## 3 0.2 0.6322188
## 4 0.3 0.7560976
## 5 0.4 0.7127072
## 6 0.5 0.7127072
## 7 0.6 0.7127072
## 8 0.7 0.7127072
## 9 0.8 0.6882353
## 10 0.9 0.6845238
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.3000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.RFE.X.rpart.N
## 1 N 390
## 2 Y 106
## sold.fctr.predict.RFE.X.rpart.Y
## 1 94
## 2 310
## Prediction
## Reference N Y
## N 390 94
## Y 106 310
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.777778e-01 5.520962e-01 7.491748e-01 8.045491e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.303387e-50 4.366766e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only## id
## 1 RFE.X.rpart
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 5 1.827 0.046
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8184969 0.7 0.7658473 0.7921358
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7850331 0.8355626 0.5756141 0.8112617
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8092816 0.3 0.7560976 0.7777778
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7491748 0.8045491 0.5520962 0.7777778
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.0157784 0.03339769
## label step_major step_minor label_minor bgn end
## 6 fit.models_1_RFE.X 2 4 rpart 138.244 143.727
## 7 fit.models_1_RFE.X 2 5 gbm 143.727 NA
## elapsed
## 6 5.483
## 7 NA
## [1] "fitting model: RFE.X.gbm"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr"## Loading required package: gbm
## Loading required package: splines
## Loaded gbm 2.1.1## Aggregating results
## Selecting tuning parameters
## Fitting n.trees = 100, interaction.depth = 5, shrinkage = 0.1, n.minobsinnode = 10 on full training set## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 22: prdl.descr.my.fctriPad5#0 has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 34: prdl.descr.my.fctriPadminiRetina#1 has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 41: cellular.fctr1:carrier.fctrNone has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 42: cellular.fctrUnknown:carrier.fctrNone has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 43: cellular.fctr0:carrier.fctrOther has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 45: cellular.fctrUnknown:carrier.fctrOther has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 46: cellular.fctr0:carrier.fctrSprint has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 48: cellular.fctrUnknown:carrier.fctrSprint has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 49: cellular.fctr0:carrier.fctrT-Mobile has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 51: cellular.fctrUnknown:carrier.fctrT-Mobile has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 52: cellular.fctr0:carrier.fctrUnknown has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 55: cellular.fctr0:carrier.fctrVerizon has no variation.## Warning in gbm.fit(x = structure(c(0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, :
## variable 57: cellular.fctrUnknown:carrier.fctrVerizon has no variation.
## Iter TrainDeviance ValidDeviance StepSize Improve
## 1 1.3031 nan 0.1000 0.0378
## 2 1.2387 nan 0.1000 0.0289
## 3 1.1817 nan 0.1000 0.0252
## 4 1.1325 nan 0.1000 0.0215
## 5 1.0939 nan 0.1000 0.0161
## 6 1.0588 nan 0.1000 0.0158
## 7 1.0321 nan 0.1000 0.0110
## 8 1.0091 nan 0.1000 0.0111
## 9 0.9895 nan 0.1000 0.0075
## 10 0.9698 nan 0.1000 0.0083
## 20 0.8611 nan 0.1000 0.0010
## 40 0.7742 nan 0.1000 -0.0015
## 60 0.7214 nan 0.1000 0.0006
## 80 0.6809 nan 0.1000 -0.0018
## 100 0.6442 nan 0.1000 -0.0010## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = mdl_id_pfx, : model's bestTune found at an extreme of
## tuneGrid for parameter: interaction.depth## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = mdl_id_pfx, : model's bestTune found at an extreme of
## tuneGrid for parameter: shrinkage## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = mdl_id_pfx, : model's bestTune found at an extreme of
## tuneGrid for parameter: n.minobsinnode
## var
## startprice startprice
## biddable biddable
## condition.fctrFor parts or not working condition.fctrFor parts or not working
## cellular.fctrUnknown cellular.fctrUnknown
## color.fctrUnknown color.fctrUnknown
## prdl.descr.my.fctriPadAir#0 prdl.descr.my.fctriPadAir#0
## cellular.fctr1 cellular.fctr1
## storage.fctr16 storage.fctr16
## prdl.descr.my.fctriPadAir2#0 prdl.descr.my.fctriPadAir2#0
## prdl.descr.my.fctriPadmini2#0 prdl.descr.my.fctriPadmini2#0
## storage.fctrUnknown storage.fctrUnknown
## prdl.descr.my.fctriPad4#0 prdl.descr.my.fctriPad4#0
## prdl.descr.my.fctriPad1#0 prdl.descr.my.fctriPad1#0
## storage.fctr64 storage.fctr64
## condition.fctrNew condition.fctrNew
## color.fctrWhite color.fctrWhite
## prdl.descr.my.fctriPad4#1 prdl.descr.my.fctriPad4#1
## prdl.descr.my.fctriPadmini#1 prdl.descr.my.fctriPadmini#1
## condition.fctrSeller refurbished condition.fctrSeller refurbished
## cellular.fctr1:carrier.fctrUnknown cellular.fctr1:carrier.fctrUnknown
## prdl.descr.my.fctriPadmini#0 prdl.descr.my.fctriPadmini#0
## prdl.descr.my.fctriPad2#1 prdl.descr.my.fctriPad2#1
## color.fctrSpace Gray color.fctrSpace Gray
## cellular.fctr1:carrier.fctrVerizon cellular.fctr1:carrier.fctrVerizon
## prdl.descr.my.fctriPad3#0 prdl.descr.my.fctriPad3#0
## prdl.descr.my.fctriPad1#1 prdl.descr.my.fctriPad1#1
## prdl.descr.my.fctrUnknown#1 prdl.descr.my.fctrUnknown#1
## prdl.descr.my.fctriPadAir#1 prdl.descr.my.fctriPadAir#1
## prdl.descr.my.fctriPadAir2#1 prdl.descr.my.fctriPadAir2#1
## prdl.descr.my.fctriPad3#1 prdl.descr.my.fctriPad3#1
## prdl.descr.my.fctriPad2#0 prdl.descr.my.fctriPad2#0
## storage.fctr32 storage.fctr32
## cellular.fctr0:carrier.fctrNone cellular.fctr0:carrier.fctrNone
## condition.fctrNew other (see details) condition.fctrNew other (see details)
## prdl.descr.my.fctriPadmini3#0 prdl.descr.my.fctriPadmini3#0
## condition.fctrManufacturer refurbished condition.fctrManufacturer refurbished
## color.fctrGold color.fctrGold
## prdl.descr.my.fctriPad5#0 prdl.descr.my.fctriPad5#0
## prdl.descr.my.fctriPadmini2#1 prdl.descr.my.fctriPadmini2#1
## prdl.descr.my.fctriPadmini3#1 prdl.descr.my.fctriPadmini3#1
## prdl.descr.my.fctriPadminiRetina#0 prdl.descr.my.fctriPadminiRetina#0
## prdl.descr.my.fctriPadminiRetina#1 prdl.descr.my.fctriPadminiRetina#1
## cellular.fctr1:carrier.fctrNone cellular.fctr1:carrier.fctrNone
## cellular.fctrUnknown:carrier.fctrNone cellular.fctrUnknown:carrier.fctrNone
## cellular.fctr0:carrier.fctrOther cellular.fctr0:carrier.fctrOther
## cellular.fctr1:carrier.fctrOther cellular.fctr1:carrier.fctrOther
## cellular.fctrUnknown:carrier.fctrOther cellular.fctrUnknown:carrier.fctrOther
## cellular.fctr0:carrier.fctrSprint cellular.fctr0:carrier.fctrSprint
## cellular.fctr1:carrier.fctrSprint cellular.fctr1:carrier.fctrSprint
## cellular.fctrUnknown:carrier.fctrSprint cellular.fctrUnknown:carrier.fctrSprint
## cellular.fctr0:carrier.fctrT-Mobile cellular.fctr0:carrier.fctrT-Mobile
## cellular.fctr1:carrier.fctrT-Mobile cellular.fctr1:carrier.fctrT-Mobile
## cellular.fctrUnknown:carrier.fctrT-Mobile cellular.fctrUnknown:carrier.fctrT-Mobile
## cellular.fctr0:carrier.fctrUnknown cellular.fctr0:carrier.fctrUnknown
## cellular.fctrUnknown:carrier.fctrUnknown cellular.fctrUnknown:carrier.fctrUnknown
## cellular.fctr0:carrier.fctrVerizon cellular.fctr0:carrier.fctrVerizon
## cellular.fctrUnknown:carrier.fctrVerizon cellular.fctrUnknown:carrier.fctrVerizon
## rel.inf
## startprice 46.11376401
## biddable 24.73060134
## condition.fctrFor parts or not working 2.73836141
## cellular.fctrUnknown 2.36114504
## color.fctrUnknown 1.60146118
## prdl.descr.my.fctriPadAir#0 1.52853619
## cellular.fctr1 1.46086616
## storage.fctr16 1.32277291
## prdl.descr.my.fctriPadAir2#0 1.23623248
## prdl.descr.my.fctriPadmini2#0 1.15288755
## storage.fctrUnknown 1.12538832
## prdl.descr.my.fctriPad4#0 1.03744553
## prdl.descr.my.fctriPad1#0 1.02337341
## storage.fctr64 0.90453384
## condition.fctrNew 0.85365474
## color.fctrWhite 0.83343812
## prdl.descr.my.fctriPad4#1 0.81880817
## prdl.descr.my.fctriPadmini#1 0.80829167
## condition.fctrSeller refurbished 0.76721021
## cellular.fctr1:carrier.fctrUnknown 0.58755084
## prdl.descr.my.fctriPadmini#0 0.58300210
## prdl.descr.my.fctriPad2#1 0.57910339
## color.fctrSpace Gray 0.57770042
## cellular.fctr1:carrier.fctrVerizon 0.54659663
## prdl.descr.my.fctriPad3#0 0.54439306
## prdl.descr.my.fctriPad1#1 0.54172357
## prdl.descr.my.fctrUnknown#1 0.48262843
## prdl.descr.my.fctriPadAir#1 0.47007372
## prdl.descr.my.fctriPadAir2#1 0.43976300
## prdl.descr.my.fctriPad3#1 0.42716371
## prdl.descr.my.fctriPad2#0 0.42510531
## storage.fctr32 0.40573118
## cellular.fctr0:carrier.fctrNone 0.36061748
## condition.fctrNew other (see details) 0.26636286
## prdl.descr.my.fctriPadmini3#0 0.14974242
## condition.fctrManufacturer refurbished 0.12892288
## color.fctrGold 0.06504671
## prdl.descr.my.fctriPad5#0 0.00000000
## prdl.descr.my.fctriPadmini2#1 0.00000000
## prdl.descr.my.fctriPadmini3#1 0.00000000
## prdl.descr.my.fctriPadminiRetina#0 0.00000000
## prdl.descr.my.fctriPadminiRetina#1 0.00000000
## cellular.fctr1:carrier.fctrNone 0.00000000
## cellular.fctrUnknown:carrier.fctrNone 0.00000000
## cellular.fctr0:carrier.fctrOther 0.00000000
## cellular.fctr1:carrier.fctrOther 0.00000000
## cellular.fctrUnknown:carrier.fctrOther 0.00000000
## cellular.fctr0:carrier.fctrSprint 0.00000000
## cellular.fctr1:carrier.fctrSprint 0.00000000
## cellular.fctrUnknown:carrier.fctrSprint 0.00000000
## cellular.fctr0:carrier.fctrT-Mobile 0.00000000
## cellular.fctr1:carrier.fctrT-Mobile 0.00000000
## cellular.fctrUnknown:carrier.fctrT-Mobile 0.00000000
## cellular.fctr0:carrier.fctrUnknown 0.00000000
## cellular.fctrUnknown:carrier.fctrUnknown 0.00000000
## cellular.fctr0:carrier.fctrVerizon 0.00000000
## cellular.fctrUnknown:carrier.fctrVerizon 0.00000000
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.7084337
## 3 0.2 0.8064516
## 4 0.3 0.8531915
## 5 0.4 0.8532423
## 6 0.5 0.8497041
## 7 0.6 0.8189763
## 8 0.7 0.7968750
## 9 0.8 0.7438017
## 10 0.9 0.6341463
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.RFE.X.gbm.N sold.fctr.predict.RFE.X.gbm.Y
## 1 N 455 60
## 2 Y 69 375
## Prediction
## Reference N Y
## N 455 60
## Y 69 375
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.654849e-01 7.291085e-01 8.422444e-01 8.864530e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 7.485894e-105 4.812082e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6649874
## 3 0.2 0.7195243
## 4 0.3 0.7248908
## 5 0.4 0.7311321
## 6 0.5 0.7279597
## 7 0.6 0.7148594
## 8 0.7 0.7028571
## 9 0.8 0.6544901
## 10 0.9 0.5704809
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.RFE.X.gbm.N sold.fctr.predict.RFE.X.gbm.Y
## 1 N 362 122
## 2 Y 106 310
## Prediction
## Reference N Y
## N 362 122
## Y 106 310
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.466667e-01 4.917974e-01 7.169179e-01 7.747908e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 2.950210e-38 3.205154e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only## id
## 1 RFE.X.gbm
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 25 7.066 0.523
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.940921 0.4 0.8532423 0.8057003
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.8422444 0.886453 0.6066585 0.8654849
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8223538 0.4 0.7311321 0.7466667
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7169179 0.7747908 0.4917974 0.7466667
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01445201 0.02952749
## label step_major step_minor label_minor bgn end
## 7 fit.models_1_RFE.X 2 5 gbm 143.727 154.863
## 8 fit.models_1_RFE.X 2 6 rf 154.863 NA
## elapsed
## 7 11.136
## 8 NA
## [1] "fitting model: RFE.X.rf"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr"## Loading required package: randomForest
## randomForest 4.6-12
## Type rfNews() to see new features/changes/bug fixes.
##
## Attaching package: 'randomForest'
##
## The following object is masked from 'package:dplyr':
##
## combine
##
## The following object is masked from 'package:gdata':
##
## combine
##
## The following object is masked from 'package:ggplot2':
##
## margin
## Aggregating results
## Selecting tuning parameters
## Fitting mtry = 43 on full training set
## Length Class Mode
## call 4 -none- call
## type 1 -none- character
## predicted 959 factor numeric
## err.rate 1500 -none- numeric
## confusion 6 -none- numeric
## votes 1918 matrix numeric
## oob.times 959 -none- numeric
## classes 2 -none- character
## importance 57 -none- numeric
## importanceSD 0 -none- NULL
## localImportance 0 -none- NULL
## proximity 0 -none- NULL
## ntree 1 -none- numeric
## mtry 1 -none- numeric
## forest 14 -none- list
## y 959 factor numeric
## test 0 -none- NULL
## inbag 0 -none- NULL
## xNames 57 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 2 -none- character
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.8222222
## 3 0.2 0.9367089
## 4 0.3 0.9768977
## 5 0.4 0.9955157
## 6 0.5 0.9988726
## 7 0.6 0.9966102
## 8 0.7 0.9415971
## 9 0.8 0.8759494
## 10 0.9 0.7802198
## 11 1.0 0.3646409## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.RFE.X.rf.N sold.fctr.predict.RFE.X.rf.Y
## 1 N 515 NA
## 2 Y 1 443
## Prediction
## Reference N Y
## N 515 0
## Y 1 443
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 9.989572e-01 9.979027e-01 9.942040e-01 9.999736e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 9.484352e-257 1.000000e+00## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6800000
## 3 0.2 0.7098646
## 4 0.3 0.7268908
## 5 0.4 0.7303754
## 6 0.5 0.7227723
## 7 0.6 0.7197875
## 8 0.7 0.6857963
## 9 0.8 0.6606335
## 10 0.9 0.5934426
## 11 1.0 0.2792608## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.RFE.X.rf.N sold.fctr.predict.RFE.X.rf.Y
## 1 N 342 142
## 2 Y 95 321
## Prediction
## Reference N Y
## N 342 142
## Y 95 321
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.366667e-01 4.744804e-01 7.065915e-01 7.651836e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.024035e-34 2.807902e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id
## 1 RFE.X.rf
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 5 9.261 3.682
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.9999978 0.5 0.9988726 0.8216893
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.994204 0.9999736 0.6393413 0.9989572
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8240549 0.4 0.7303754 0.7366667
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7065915 0.7651836 0.4744804 0.7366667
## label step_major step_minor label_minor bgn end
## 8 fit.models_1_RFE.X 2 6 rf 154.863 167.584
## 9 fit.models_1_All.X 3 0 setup 167.585 NA
## elapsed
## 8 12.721
## 9 NA
## label step_major step_minor label_minor bgn end
## 9 fit.models_1_All.X 3 0 setup 167.585 167.591
## 10 fit.models_1_All.X 3 1 glmnet 167.592 NA
## elapsed
## 9 0.006
## 10 NA
## [1] "fitting model: All.X.glmnet"
## [1] " indep_vars: biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice,cellular.fctr:carrier.fctr"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.55, lambda = 0.0252 on full training set
## Length Class Mode
## a0 86 -none- numeric
## beta 4988 dgCMatrix S4
## df 86 -none- numeric
## dim 2 -none- numeric
## lambda 86 -none- numeric
## dev.ratio 86 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 58 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept)
## 0.212280420
## biddable
## 1.517695590
## cellular.fctr1
## 0.111291692
## cellular.fctrUnknown
## -0.240821550
## condition.fctrSeller refurbished
## -0.303613584
## prdl.descr.my.fctrUnknown#1
## -0.001428038
## prdl.descr.my.fctriPad2#0
## 0.578460972
## prdl.descr.my.fctriPad2#1
## -0.138431499
## prdl.descr.my.fctriPad3#0
## 0.193090727
## prdl.descr.my.fctriPad4#0
## 0.239392597
## prdl.descr.my.fctriPad4#1
## -0.926942426
## prdl.descr.my.fctriPadAir#0
## 0.356239530
## prdl.descr.my.fctriPadAir2#0
## 0.496926272
## prdl.descr.my.fctriPadAir2#1
## 0.329623264
## prdl.descr.my.fctriPadmini#0
## -0.054182251
## prdl.descr.my.fctriPadmini2#0
## 0.071308286
## prdl.descr.my.fctriPadmini3#0
## -0.058991667
## prdl.descr.my.fctriPadminiRetina#0
## 0.615538025
## startprice
## -0.005378116
## storage.fctr32
## -0.012610177
## storage.fctr64
## 0.062650081
## cellular.fctr1:carrier.fctrT-Mobile
## 0.198632683
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.232792402
## cellular.fctr1:carrier.fctrVerizon
## 0.039311991
## [1] "max lambda < lambdaOpt:"
## (Intercept)
## 0.242008407
## biddable
## 1.519615706
## cellular.fctr1
## 0.118581035
## cellular.fctrUnknown
## -0.251456870
## condition.fctrSeller refurbished
## -0.341299917
## prdl.descr.my.fctrUnknown#1
## -0.016706048
## prdl.descr.my.fctriPad2#0
## 0.620952079
## prdl.descr.my.fctriPad2#1
## -0.167333182
## prdl.descr.my.fctriPad3#0
## 0.225106750
## prdl.descr.my.fctriPad4#0
## 0.288944274
## prdl.descr.my.fctriPad4#1
## -0.963264406
## prdl.descr.my.fctriPadAir#0
## 0.426395470
## prdl.descr.my.fctriPadAir2#0
## 0.587153313
## prdl.descr.my.fctriPadAir2#1
## 0.421222039
## prdl.descr.my.fctriPadmini#0
## -0.066928131
## prdl.descr.my.fctriPadmini2#0
## 0.133141513
## prdl.descr.my.fctriPadmini3#0
## -0.059558089
## prdl.descr.my.fctriPadminiRetina#0
## 0.775062295
## startprice
## -0.005612763
## storage.fctr32
## -0.029076706
## storage.fctr64
## 0.078190707
## cellular.fctr1:carrier.fctrT-Mobile
## 0.294925866
## cellular.fctr1:carrier.fctrUnknown
## 0.024140616
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.241582561
## cellular.fctr1:carrier.fctrVerizon
## 0.069556614
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6666667
## 3 0.2 0.7179487
## 4 0.3 0.7579556
## 5 0.4 0.7662338
## 6 0.5 0.7792208
## 7 0.6 0.7660099
## 8 0.7 0.7591623
## 9 0.8 0.5931677
## 10 0.9 0.0944206
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.All.X.glmnet.N
## 1 N 442
## 2 Y 114
## sold.fctr.predict.All.X.glmnet.Y
## 1 73
## 2 330
## Prediction
## Reference N Y
## N 442 73
## Y 114 330
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.050052e-01 6.053489e-01 7.784873e-01 8.296347e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 1.430019e-67 3.443570e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.63221884
## 2 0.1 0.65329053
## 3 0.2 0.68177697
## 4 0.3 0.70647773
## 5 0.4 0.73198198
## 6 0.5 0.75735294
## 7 0.6 0.73850197
## 8 0.7 0.69132290
## 9 0.8 0.51027397
## 10 0.9 0.08715596
## 11 1.0 0.00000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.All.X.glmnet.N
## 1 N 393
## 2 Y 107
## sold.fctr.predict.All.X.glmnet.Y
## 1 91
## 2 309
## Prediction
## Reference N Y
## N 393 91
## Y 107 309
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.800000e-01 5.562749e-01 7.514869e-01 8.066667e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 1.417978e-51 2.864220e-01## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id
## 1 All.X.glmnet
## feats
## 1 biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 25 2.81 0.095
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8767777 0.5 0.7792208 0.7949168
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7784873 0.8296347 0.5850163 0.8050052
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8178863 0.5 0.7573529 0.78
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7514869 0.8066667 0.5562749 0.78
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.008376313 0.01748681
## label step_major step_minor label_minor bgn
## 10 fit.models_1_All.X 3 1 glmnet 167.592
## 11 fit.models_1_Best.Interact 4 0 setup 174.376
## end elapsed
## 10 174.375 6.783
## 11 NA NA
## label step_major step_minor
## 11 fit.models_1_Best.Interact 4 0
## 12 fit.models_1_Max.cor.Y.rcv.3X1.Interact 4 1
## label_minor bgn end elapsed
## 11 setup 174.376 174.401 0.025
## 12 glmnet 174.402 NA NA
## [1] "fitting model: Max.cor.Y.rcv.3X1.Interact.glmnet"
## [1] " indep_vars: startprice,biddable*"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.325, lambda = 0.117 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst
## = list(id.prefix = mdl_id_pfx, : model's bestTune found at an extreme of
## tuneGrid for parameter: lambda
## Length Class Mode
## a0 73 -none- numeric
## beta 219 dgCMatrix S4
## df 73 -none- numeric
## dim 2 -none- numeric
## lambda 73 -none- numeric
## dev.ratio 73 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 3 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept) biddable startprice
## -0.067691308 1.172894109 -0.003029053
## [1] "max lambda < lambdaOpt:"
## (Intercept) biddable startprice
## -0.059896920 1.215957253 -0.003170641
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6329294
## 2 0.1 0.6374731
## 3 0.2 0.6631179
## 4 0.3 0.7170475
## 5 0.4 0.7636364
## 6 0.5 0.7603306
## 7 0.6 0.7450000
## 8 0.7 0.6219512
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.4000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X1.Interact.glmnet.N
## 1 N 381
## 2 Y 87
## sold.fctr.predict.Max.cor.Y.rcv.3X1.Interact.glmnet.Y
## 1 134
## 2 357
## Prediction
## Reference N Y
## N 381 134
## Y 87 357
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.695516e-01 5.399202e-01 7.415720e-01 7.958649e-01 5.370177e-01
## AccuracyPValue McnemarPValue
## 2.083514e-50 1.972813e-03## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## threshold f.score
## 1 0.0 0.6322188
## 2 0.1 0.6370597
## 3 0.2 0.6558966
## 4 0.3 0.6920223
## 5 0.4 0.7357631
## 6 0.5 0.7522013
## 7 0.6 0.7272727
## 8 0.7 0.5623960
## 9 0.8 0.0000000
## 10 0.9 0.0000000
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.5000 to maximize f.score.OOB"
## sold.fctr sold.fctr.predict.Max.cor.Y.rcv.3X1.Interact.glmnet.N
## 1 N 404
## 2 Y 117
## sold.fctr.predict.Max.cor.Y.rcv.3X1.Interact.glmnet.Y
## 1 80
## 2 299
## Prediction
## Reference N Y
## N 404 80
## Y 117 299
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 7.811111e-01 5.569405e-01 7.526434e-01 8.077250e-01 5.377778e-01
## AccuracyPValue McnemarPValue
## 4.632073e-52 1.032074e-02## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## id feats max.nTuningRuns
## 1 Max.cor.Y.rcv.3X1.Interact.glmnet startprice,biddable* 25
## min.elapsedtime.everything min.elapsedtime.final max.auc.fit
## 1 2.17 0.019 0.8568748
## opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.4 0.7636364 0.7872682
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.741572 0.7958649 0.5694675 0.7695516
## max.auc.OOB opt.prob.threshold.OOB max.f.score.OOB max.Accuracy.OOB
## 1 0.8262948 0.5 0.7522013 0.7811111
## max.AccuracyLower.OOB max.AccuracyUpper.OOB max.Kappa.OOB .OOB
## 1 0.7526434 0.807725 0.5569405 0.7811111
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01250943 0.02544908# Check if other preProcess methods improve model performance
mdl_id <- orderBy(get_model_sel_frmla(), glb_models_df)[1, "id"]
indep_vars_vctr <-
trim(unlist(strsplit(glb_models_df[glb_models_df$id == mdl_id, "feats"], "[,]")))
method <- tail(unlist(strsplit(mdl_id, "[.]")), 1)
mdl_id_pfx <- paste0(head(unlist(strsplit(mdl_id, "[.]")), -1), collapse=".")
for (prePr in glb_preproc_methods) {
# The operations are applied in this order:
# Box-Cox/Yeo-Johnson transformation, centering, scaling, range, imputation, PCA, ICA then spatial sign.
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix=mdl_id_pfx,
type=glb_model_type, tune.df=glb_tune_models_df,
trainControl.method="repeatedcv",
trainControl.number=glb_rcv_n_folds, trainControl.repeats=glb_rcv_n_repeats,
train.method=method, train.preProcess=prePr)),
indep_vars=indep_vars_vctr, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)
}
# If (All|RFE).X.glm is less accurate than Low.Cor.X.glm
# check NA coefficients & filter appropriate terms in indep_vars_vctr
# if (method == "glm") {
# orig_glm <- glb_models_lst[[paste0(mdl_id, ".", model_method)]]$finalModel
# orig_glm <- glb_models_lst[["All.X.glm"]]$finalModel; print(summary(orig_glm))
# orig_glm <- glb_models_lst[["RFE.X.glm"]]$finalModel; print(summary(orig_glm))
# require(car)
# vif_orig_glm <- vif(orig_glm); print(vif_orig_glm)
# # if vif errors out with "there are aliased coefficients in the model"
# alias_orig_glm <- alias(orig_glm); alias_complete_orig_glm <- (alias_orig_glm$Complete > 0); alias_complete_orig_glm <- alias_complete_orig_glm[rowSums(alias_complete_orig_glm) > 0, colSums(alias_complete_orig_glm) > 0]; print(alias_complete_orig_glm)
# print(vif_orig_glm[!is.na(vif_orig_glm) & (vif_orig_glm == Inf)])
# print(which.max(vif_orig_glm))
# print(sort(vif_orig_glm[vif_orig_glm >= 1.0e+03], decreasing=TRUE))
# glb_fitobs_df[c(1143, 3637, 3953, 4105), c("UniqueID", "Popular", "H.P.quandary", "Headline")]
# glb_feats_df[glb_feats_df$id %in% grep("[HSA]\\.chrs.n.log", glb_feats_df$id, value=TRUE) | glb_feats_df$cor.high.X %in% grep("[HSA]\\.chrs.n.log", glb_feats_df$id, value=TRUE), ]
# all.equal(glb_allobs_df$S.chrs.uppr.n.log, glb_allobs_df$A.chrs.uppr.n.log)
# cor(glb_allobs_df$S.T.herald, glb_allobs_df$S.T.tribun)
# mydsp_obs(Abstract.contains="[Dd]iar", cols=("Abstract"), all=TRUE)
# subset(glb_feats_df, cor.y.abs <= glb_feats_df[glb_feats_df$id == ".rnorm", "cor.y.abs"])
# corxx_mtrx <- cor(data.matrix(glb_allobs_df[, setdiff(names(glb_allobs_df), myfind_chr_cols_df(glb_allobs_df))]), use="pairwise.complete.obs"); abs_corxx_mtrx <- abs(corxx_mtrx); diag(abs_corxx_mtrx) <- 0
# which.max(abs_corxx_mtrx["S.T.tribun", ])
# abs_corxx_mtrx["A.npnct08.log", "S.npnct08.log"]
# step_glm <- step(orig_glm)
# }
# Since caret does not optimize rpart well
# if (method == "rpart")
# ret_lst <- myfit_mdl(mdl_id=paste0(mdl_id_pfx, ".cp.0"), model_method=method,
# indep_vars_vctr=indep_vars_vctr,
# model_type=glb_model_type,
# rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
# fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df,
# n_cv_folds=0, tune_models_df=data.frame(parameter="cp", min=0.0, max=0.0, by=0.1))
# User specified
# Ensure at least 2 vars in each regression; else varImp crashes
# sav_models_lst <- glb_models_lst; sav_models_df <- glb_models_df; sav_featsimp_df <- glb_featsimp_df; all.equal(sav_featsimp_df, glb_featsimp_df)
# glb_models_lst <- sav_models_lst; glb_models_df <- sav_models_df; glm_featsimp_df <- sav_featsimp_df
# easier to exclude features
# require(gdata) # needed for trim
# mdl_id <- "";
# indep_vars_vctr <- head(subset(glb_models_df, grepl("All\\.X\\.", mdl_id), select=feats)
# , 1)[, "feats"]
# indep_vars_vctr <- trim(unlist(strsplit(indep_vars_vctr, "[,]")))
# indep_vars_vctr <- setdiff(indep_vars_vctr, ".rnorm")
# easier to include features
#stop(here"); sav_models_df <- glb_models_df; glb_models_df <- sav_models_df
# !_sp
# mdl_id <- "csm"; indep_vars_vctr <- c(NULL
# ,"prdline.my.fctr", "prdline.my.fctr:.clusterid.fctr"
# ,"prdline.my.fctr*biddable"
# #,"prdline.my.fctr*startprice.log"
# #,"prdline.my.fctr*startprice.diff"
# ,"prdline.my.fctr*condition.fctr"
# ,"prdline.my.fctr*D.terms.post.stop.n"
# #,"prdline.my.fctr*D.terms.post.stem.n"
# ,"prdline.my.fctr*cellular.fctr"
# # ,"<feat1>:<feat2>"
# )
# for (method in glb_mdl_methods) {
# ret_lst <- myfit_mdl(mdl_id=mdl_id, model_method=method,
# indep_vars_vctr=indep_vars_vctr,
# model_type=glb_model_type,
# rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
# fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df,
# n_cv_folds=glb_rcv_n_folds, tune_models_df=glb_tune_models_df)
# csm_mdl_id <- paste0(mdl_id, ".", method)
# csm_featsimp_df <- myget_feats_importance(glb_models_lst[[paste0(mdl_id, ".",
# method)]]); print(head(csm_featsimp_df))
# }
###
# Ntv.1.lm <- lm(reformulate(indep_vars_vctr, glb_rsp_var), glb_trnobs_df); print(summary(Ntv.1.lm))
#glb_models_df[, "max.Accuracy.OOB", FALSE]
#varImp(glb_models_lst[["Low.cor.X.glm"]])
#orderBy(~ -Overall, varImp(glb_models_lst[["All.X.2.glm"]])$importance)
#orderBy(~ -Overall, varImp(glb_models_lst[["All.X.3.glm"]])$importance)
#glb_feats_df[grepl("npnct28", glb_feats_df$id), ]
# User specified bivariate models
# indep_vars_vctr_lst <- list()
# for (feat in setdiff(names(glb_fitobs_df),
# union(glb_rsp_var, glb_exclude_vars_as_features)))
# indep_vars_vctr_lst[["feat"]] <- feat
# User specified combinatorial models
# indep_vars_vctr_lst <- list()
# combn_mtrx <- combn(c("<feat1_name>", "<feat2_name>", "<featn_name>"),
# <num_feats_to_choose>)
# for (combn_ix in 1:ncol(combn_mtrx))
# #print(combn_mtrx[, combn_ix])
# indep_vars_vctr_lst[[combn_ix]] <- combn_mtrx[, combn_ix]
# template for myfit_mdl
# rf is hard-coded in caret to recognize only Accuracy / Kappa evaluation metrics
# only for OOB in trainControl ?
# ret_lst <- myfit_mdl_fn(mdl_id=paste0(mdl_id_pfx, ""), model_method=method,
# indep_vars_vctr=indep_vars_vctr,
# rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
# fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df,
# n_cv_folds=glb_rcv_n_folds, tune_models_df=glb_tune_models_df,
# model_loss_mtrx=glb_model_metric_terms,
# model_summaryFunction=glb_model_metric_smmry,
# model_metric=glb_model_metric,
# model_metric_maximize=glb_model_metric_maximize)
# Simplify a model
# fit_df <- glb_fitobs_df; glb_mdl <- step(<complex>_mdl)
# Non-caret models
# rpart_area_mdl <- rpart(reformulate("Area", response=glb_rsp_var),
# data=glb_fitobs_df, #method="class",
# control=rpart.control(cp=0.12),
# parms=list(loss=glb_model_metric_terms))
# print("rpart_sel_wlm_mdl"); prp(rpart_sel_wlm_mdl)
#
print(glb_models_df)## id
## MFO.myMFO_classfr MFO.myMFO_classfr
## Random.myrandom_classfr Random.myrandom_classfr
## Max.cor.Y.rcv.1X1.glmnet Max.cor.Y.rcv.1X1.glmnet
## Max.cor.Y.rcv.3X1.glmnet Max.cor.Y.rcv.3X1.glmnet
## Max.cor.Y.rcv.3X3.glmnet Max.cor.Y.rcv.3X3.glmnet
## Max.cor.Y.rcv.3X5.glmnet Max.cor.Y.rcv.3X5.glmnet
## Max.cor.Y.rcv.5X1.glmnet Max.cor.Y.rcv.5X1.glmnet
## Max.cor.Y.rcv.5X3.glmnet Max.cor.Y.rcv.5X3.glmnet
## Max.cor.Y.rcv.5X5.glmnet Max.cor.Y.rcv.5X5.glmnet
## Max.cor.Y.rcv.1X1.cp.0.rpart Max.cor.Y.rcv.1X1.cp.0.rpart
## Max.cor.Y.rpart Max.cor.Y.rpart
## Interact.High.cor.Y.glmnet Interact.High.cor.Y.glmnet
## Low.cor.X.glmnet Low.cor.X.glmnet
## RFE.X.glm RFE.X.glm
## RFE.X.bayesglm RFE.X.bayesglm
## RFE.X.glmnet RFE.X.glmnet
## RFE.X.rpart RFE.X.rpart
## RFE.X.gbm RFE.X.gbm
## RFE.X.rf RFE.X.rf
## All.X.glmnet All.X.glmnet
## Max.cor.Y.rcv.3X1.Interact.glmnet Max.cor.Y.rcv.3X1.Interact.glmnet
## feats
## MFO.myMFO_classfr .rnorm
## Random.myrandom_classfr .rnorm
## Max.cor.Y.rcv.1X1.glmnet biddable,startprice
## Max.cor.Y.rcv.3X1.glmnet biddable,startprice
## Max.cor.Y.rcv.3X3.glmnet biddable,startprice
## Max.cor.Y.rcv.3X5.glmnet biddable,startprice
## Max.cor.Y.rcv.5X1.glmnet biddable,startprice
## Max.cor.Y.rcv.5X3.glmnet biddable,startprice
## Max.cor.Y.rcv.5X5.glmnet biddable,startprice
## Max.cor.Y.rcv.1X1.cp.0.rpart biddable,startprice
## Max.cor.Y.rpart biddable,startprice
## Interact.High.cor.Y.glmnet biddable,startprice,biddable:cellular.fctr
## Low.cor.X.glmnet biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice
## RFE.X.glm startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.bayesglm startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.glmnet startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.rpart startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.gbm startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.rf startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## All.X.glmnet biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice,cellular.fctr:carrier.fctr
## Max.cor.Y.rcv.3X1.Interact.glmnet startprice,biddable*
## max.nTuningRuns
## MFO.myMFO_classfr 0
## Random.myrandom_classfr 0
## Max.cor.Y.rcv.1X1.glmnet 0
## Max.cor.Y.rcv.3X1.glmnet 25
## Max.cor.Y.rcv.3X3.glmnet 25
## Max.cor.Y.rcv.3X5.glmnet 25
## Max.cor.Y.rcv.5X1.glmnet 25
## Max.cor.Y.rcv.5X3.glmnet 25
## Max.cor.Y.rcv.5X5.glmnet 25
## Max.cor.Y.rcv.1X1.cp.0.rpart 0
## Max.cor.Y.rpart 5
## Interact.High.cor.Y.glmnet 25
## Low.cor.X.glmnet 25
## RFE.X.glm 1
## RFE.X.bayesglm 1
## RFE.X.glmnet 25
## RFE.X.rpart 5
## RFE.X.gbm 25
## RFE.X.rf 5
## All.X.glmnet 25
## Max.cor.Y.rcv.3X1.Interact.glmnet 25
## min.elapsedtime.everything
## MFO.myMFO_classfr 0.265
## Random.myrandom_classfr 0.253
## Max.cor.Y.rcv.1X1.glmnet 0.650
## Max.cor.Y.rcv.3X1.glmnet 1.367
## Max.cor.Y.rcv.3X3.glmnet 1.834
## Max.cor.Y.rcv.3X5.glmnet 2.335
## Max.cor.Y.rcv.5X1.glmnet 1.558
## Max.cor.Y.rcv.5X3.glmnet 2.252
## Max.cor.Y.rcv.5X5.glmnet 2.869
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.693
## Max.cor.Y.rpart 1.605
## Interact.High.cor.Y.glmnet 2.219
## Low.cor.X.glmnet 2.596
## RFE.X.glm 1.844
## RFE.X.bayesglm 2.819
## RFE.X.glmnet 2.675
## RFE.X.rpart 1.827
## RFE.X.gbm 7.066
## RFE.X.rf 9.261
## All.X.glmnet 2.810
## Max.cor.Y.rcv.3X1.Interact.glmnet 2.170
## min.elapsedtime.final max.auc.fit
## MFO.myMFO_classfr 0.003 0.5000000
## Random.myrandom_classfr 0.002 0.4950888
## Max.cor.Y.rcv.1X1.glmnet 0.020 0.8570235
## Max.cor.Y.rcv.3X1.glmnet 0.017 0.8560789
## Max.cor.Y.rcv.3X3.glmnet 0.016 0.8568573
## Max.cor.Y.rcv.3X5.glmnet 0.017 0.8568573
## Max.cor.Y.rcv.5X1.glmnet 0.014 0.8560789
## Max.cor.Y.rcv.5X3.glmnet 0.017 0.8568573
## Max.cor.Y.rcv.5X5.glmnet 0.016 0.8568573
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.012 0.8939014
## Max.cor.Y.rpart 0.012 0.8162184
## Interact.High.cor.Y.glmnet 0.016 0.8567917
## Low.cor.X.glmnet 0.083 0.8903306
## RFE.X.glm 0.087 0.8930683
## RFE.X.bayesglm 0.125 0.8919881
## RFE.X.glmnet 0.096 0.8767777
## RFE.X.rpart 0.046 0.8184969
## RFE.X.gbm 0.523 0.9409210
## RFE.X.rf 3.682 0.9999978
## All.X.glmnet 0.095 0.8767777
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.019 0.8568748
## opt.prob.threshold.fit max.f.score.fit
## MFO.myMFO_classfr 0.5 0.0000000
## Random.myrandom_classfr 0.4 0.6329294
## Max.cor.Y.rcv.1X1.glmnet 0.4 0.7714286
## Max.cor.Y.rcv.3X1.glmnet 0.4 0.7600849
## Max.cor.Y.rcv.3X3.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.3X5.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.5X1.glmnet 0.4 0.7600849
## Max.cor.Y.rcv.5X3.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.5X5.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.4 0.8125677
## Max.cor.Y.rpart 0.6 0.7541401
## Interact.High.cor.Y.glmnet 0.4 0.7617021
## Low.cor.X.glmnet 0.5 0.8013937
## RFE.X.glm 0.6 0.8044010
## RFE.X.bayesglm 0.5 0.7972028
## RFE.X.glmnet 0.5 0.7792208
## RFE.X.rpart 0.7 0.7658473
## RFE.X.gbm 0.4 0.8532423
## RFE.X.rf 0.5 0.9988726
## All.X.glmnet 0.5 0.7792208
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.4 0.7636364
## max.Accuracy.fit max.AccuracyLower.fit
## MFO.myMFO_classfr 0.5370177 0.5048652
## Random.myrandom_classfr 0.4629823 0.4310577
## Max.cor.Y.rcv.1X1.glmnet 0.7831074 0.7556528
## Max.cor.Y.rcv.3X1.glmnet 0.7883033 0.7361668
## Max.cor.Y.rcv.3X3.glmnet 0.7876154 0.7437357
## Max.cor.Y.rcv.3X5.glmnet 0.7879043 0.7437357
## Max.cor.Y.rcv.5X1.glmnet 0.7872982 0.7361668
## Max.cor.Y.rcv.5X3.glmnet 0.7879781 0.7437357
## Max.cor.Y.rcv.5X5.glmnet 0.7874989 0.7437357
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.8196038 0.7937768
## Max.cor.Y.rpart 0.7824049 0.7719514
## Interact.High.cor.Y.glmnet 0.7872682 0.7383282
## Low.cor.X.glmnet 0.7942268 0.7959657
## RFE.X.glm 0.7900546 0.8080269
## RFE.X.bayesglm 0.7904040 0.7926828
## RFE.X.glmnet 0.7949168 0.7784873
## RFE.X.rpart 0.7921358 0.7850331
## RFE.X.gbm 0.8057003 0.8422444
## RFE.X.rf 0.8216893 0.9942040
## All.X.glmnet 0.7949168 0.7784873
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7872682 0.7415720
## max.AccuracyUpper.fit max.Kappa.fit
## MFO.myMFO_classfr 0.5689423 0.0000000
## Random.myrandom_classfr 0.4951348 0.0000000
## Max.cor.Y.rcv.1X1.glmnet 0.8088110 0.5653087
## Max.cor.Y.rcv.3X1.glmnet 0.7908752 0.5717602
## Max.cor.Y.rcv.3X3.glmnet 0.7978592 0.5701499
## Max.cor.Y.rcv.3X5.glmnet 0.7978592 0.5707133
## Max.cor.Y.rcv.5X1.glmnet 0.7908752 0.5698703
## Max.cor.Y.rcv.5X3.glmnet 0.7978592 0.5707928
## Max.cor.Y.rcv.5X5.glmnet 0.7978592 0.5698279
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.8434503 0.6391797
## Max.cor.Y.rpart 0.8236969 0.5580889
## Interact.High.cor.Y.glmnet 0.7928718 0.5695577
## Low.cor.X.glmnet 0.8454193 0.5846679
## RFE.X.glm 0.8562261 0.5764108
## RFE.X.bayesglm 0.8424654 0.5769613
## RFE.X.glmnet 0.8296347 0.5850163
## RFE.X.rpart 0.8355626 0.5756141
## RFE.X.gbm 0.8864530 0.6066585
## RFE.X.rf 0.9999736 0.6393413
## All.X.glmnet 0.8296347 0.5850163
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7958649 0.5694675
## .fit max.auc.OOB
## MFO.myMFO_classfr 0.5370177 0.5000000
## Random.myrandom_classfr 0.4629823 0.5162111
## Max.cor.Y.rcv.1X1.glmnet 0.7831074 0.8253362
## Max.cor.Y.rcv.3X1.glmnet 0.7643379 0.8266524
## Max.cor.Y.rcv.3X3.glmnet 0.7716371 0.8260961
## Max.cor.Y.rcv.3X5.glmnet 0.7716371 0.8260961
## Max.cor.Y.rcv.5X1.glmnet 0.7643379 0.8266524
## Max.cor.Y.rcv.5X3.glmnet 0.7716371 0.8260961
## Max.cor.Y.rcv.5X5.glmnet 0.7716371 0.8260961
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.8196038 0.7971655
## Max.cor.Y.rpart 0.7987487 0.8095598
## Interact.High.cor.Y.glmnet 0.7664234 0.8262948
## Low.cor.X.glmnet 0.8216893 0.8131655
## RFE.X.glm 0.8331595 0.8055666
## RFE.X.bayesglm 0.8185610 0.8111689
## RFE.X.glmnet 0.8050052 0.8178863
## RFE.X.rpart 0.8112617 0.8092816
## RFE.X.gbm 0.8654849 0.8223538
## RFE.X.rf 0.9989572 0.8240549
## All.X.glmnet 0.8050052 0.8178863
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7695516 0.8262948
## opt.prob.threshold.OOB max.f.score.OOB
## MFO.myMFO_classfr 0.5 0.0000000
## Random.myrandom_classfr 0.4 0.6322188
## Max.cor.Y.rcv.1X1.glmnet 0.5 0.7474747
## Max.cor.Y.rcv.3X1.glmnet 0.5 0.7568922
## Max.cor.Y.rcv.3X3.glmnet 0.5 0.7522013
## Max.cor.Y.rcv.3X5.glmnet 0.5 0.7522013
## Max.cor.Y.rcv.5X1.glmnet 0.5 0.7568922
## Max.cor.Y.rcv.5X3.glmnet 0.5 0.7522013
## Max.cor.Y.rcv.5X5.glmnet 0.5 0.7522013
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.3 0.7260870
## Max.cor.Y.rpart 0.3 0.7560976
## Interact.High.cor.Y.glmnet 0.5 0.7522013
## Low.cor.X.glmnet 0.5 0.7436209
## RFE.X.glm 0.5 0.7406514
## RFE.X.bayesglm 0.5 0.7442424
## RFE.X.glmnet 0.5 0.7573529
## RFE.X.rpart 0.3 0.7560976
## RFE.X.gbm 0.4 0.7311321
## RFE.X.rf 0.4 0.7303754
## All.X.glmnet 0.5 0.7573529
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.5 0.7522013
## max.Accuracy.OOB max.AccuracyLower.OOB
## MFO.myMFO_classfr 0.5377778 0.5045709
## Random.myrandom_classfr 0.4622222 0.4292634
## Max.cor.Y.rcv.1X1.glmnet 0.7777778 0.7491748
## Max.cor.Y.rcv.3X1.glmnet 0.7844444 0.7561144
## Max.cor.Y.rcv.3X3.glmnet 0.7811111 0.7526434
## Max.cor.Y.rcv.3X5.glmnet 0.7811111 0.7526434
## Max.cor.Y.rcv.5X1.glmnet 0.7844444 0.7561144
## Max.cor.Y.rcv.5X3.glmnet 0.7811111 0.7526434
## Max.cor.Y.rcv.5X5.glmnet 0.7811111 0.7526434
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.7200000 0.6894226
## Max.cor.Y.rpart 0.7777778 0.7491748
## Interact.High.cor.Y.glmnet 0.7811111 0.7526434
## Low.cor.X.glmnet 0.7655556 0.7364780
## RFE.X.glm 0.7611111 0.7318690
## RFE.X.bayesglm 0.7655556 0.7364780
## RFE.X.glmnet 0.7800000 0.7514869
## RFE.X.rpart 0.7777778 0.7491748
## RFE.X.gbm 0.7466667 0.7169179
## RFE.X.rf 0.7366667 0.7065915
## All.X.glmnet 0.7800000 0.7514869
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7811111 0.7526434
## max.AccuracyUpper.OOB max.Kappa.OOB
## MFO.myMFO_classfr 0.5707366 0.0000000
## Random.myrandom_classfr 0.4954291 0.0000000
## Max.cor.Y.rcv.1X1.glmnet 0.8045491 0.5499640
## Max.cor.Y.rcv.3X1.glmnet 0.8108984 0.5639099
## Max.cor.Y.rcv.3X3.glmnet 0.8077250 0.5569405
## Max.cor.Y.rcv.3X5.glmnet 0.8077250 0.5569405
## Max.cor.Y.rcv.5X1.glmnet 0.8108984 0.5639099
## Max.cor.Y.rcv.5X3.glmnet 0.8077250 0.5569405
## Max.cor.Y.rcv.5X5.glmnet 0.8077250 0.5569405
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.7491296 0.4450317
## Max.cor.Y.rpart 0.8045491 0.5520962
## Interact.High.cor.Y.glmnet 0.8077250 0.5569405
## Low.cor.X.glmnet 0.7928829 0.5277012
## RFE.X.glm 0.7886326 0.5192355
## RFE.X.bayesglm 0.7928829 0.5278609
## RFE.X.glmnet 0.8066667 0.5562749
## RFE.X.rpart 0.8045491 0.5520962
## RFE.X.gbm 0.7747908 0.4917974
## RFE.X.rf 0.7651836 0.4744804
## All.X.glmnet 0.8066667 0.5562749
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.8077250 0.5569405
## .OOB max.AccuracySD.fit
## MFO.myMFO_classfr 0.5377778 NA
## Random.myrandom_classfr 0.4622222 NA
## Max.cor.Y.rcv.1X1.glmnet 0.7777778 NA
## Max.cor.Y.rcv.3X1.glmnet 0.7844444 0.017558849
## Max.cor.Y.rcv.3X3.glmnet 0.7811111 0.012559941
## Max.cor.Y.rcv.3X5.glmnet 0.7811111 0.016551996
## Max.cor.Y.rcv.5X1.glmnet 0.7844444 0.017147557
## Max.cor.Y.rcv.5X3.glmnet 0.7811111 0.020677417
## Max.cor.Y.rcv.5X5.glmnet 0.7811111 0.029529064
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.7200000 NA
## Max.cor.Y.rpart 0.7777778 0.013132721
## Interact.High.cor.Y.glmnet 0.7811111 0.012509433
## Low.cor.X.glmnet 0.7655556 0.019855515
## RFE.X.glm 0.7611111 0.015728416
## RFE.X.bayesglm 0.7655556 0.018255372
## RFE.X.glmnet 0.7800000 0.008376313
## RFE.X.rpart 0.7777778 0.015778402
## RFE.X.gbm 0.7466667 0.014452010
## RFE.X.rf 0.7366667 NA
## All.X.glmnet 0.7800000 0.008376313
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7811111 0.012509433
## max.KappaSD.fit
## MFO.myMFO_classfr NA
## Random.myrandom_classfr NA
## Max.cor.Y.rcv.1X1.glmnet NA
## Max.cor.Y.rcv.3X1.glmnet 0.03350704
## Max.cor.Y.rcv.3X3.glmnet 0.02551235
## Max.cor.Y.rcv.3X5.glmnet 0.03392463
## Max.cor.Y.rcv.5X1.glmnet 0.03415616
## Max.cor.Y.rcv.5X3.glmnet 0.04250265
## Max.cor.Y.rcv.5X5.glmnet 0.06009372
## Max.cor.Y.rcv.1X1.cp.0.rpart NA
## Max.cor.Y.rpart 0.02681447
## Interact.High.cor.Y.glmnet 0.02540170
## Low.cor.X.glmnet 0.03967914
## RFE.X.glm 0.03141904
## RFE.X.bayesglm 0.03653979
## RFE.X.glmnet 0.01748681
## RFE.X.rpart 0.03339769
## RFE.X.gbm 0.02952749
## RFE.X.rf NA
## All.X.glmnet 0.01748681
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.02544908rm(ret_lst)
fit.models_1_chunk_df <- myadd_chunk(fit.models_1_chunk_df, "fit.models_1_end",
major.inc=TRUE, label.minor="teardown")## label step_major step_minor
## 12 fit.models_1_Max.cor.Y.rcv.3X1.Interact 4 1
## 13 fit.models_1_end 5 0
## label_minor bgn end elapsed
## 12 glmnet 174.402 180.203 5.801
## 13 teardown 180.204 NA NAglb_chunks_df <- myadd_chunk(glb_chunks_df, "fit.models", major.inc=FALSE)## label step_major step_minor label_minor bgn end elapsed
## 11 fit.models 7 1 1 110.358 180.213 69.855
## 12 fit.models 7 2 2 180.214 NA NAif (!is.null(glb_model_metric_smmry)) {
stats_df <- glb_models_df[, "id", FALSE]
stats_mdl_df <- data.frame()
for (mdl_id in stats_df$id) {
stats_mdl_df <- rbind(stats_mdl_df,
mypredict_mdl(glb_models_lst[[mdl_id]], glb_fitobs_df, glb_rsp_var,
glb_rsp_var_out, mdl_id, "fit",
glb_model_metric_smmry, glb_model_metric,
glb_model_metric_maximize, ret_type="stats"))
}
stats_df <- merge(stats_df, stats_mdl_df, all.x=TRUE)
stats_mdl_df <- data.frame()
for (mdl_id in stats_df$id) {
stats_mdl_df <- rbind(stats_mdl_df,
mypredict_mdl(glb_models_lst[[mdl_id]], glb_OOBobs_df, glb_rsp_var,
glb_rsp_var_out, mdl_id, "OOB",
glb_model_metric_smmry, glb_model_metric,
glb_model_metric_maximize, ret_type="stats"))
}
stats_df <- merge(stats_df, stats_mdl_df, all.x=TRUE)
print("Merging following data into glb_models_df:")
print(stats_mrg_df <- stats_df[, c(1, grep(glb_model_metric, names(stats_df)))])
print(tmp_models_df <- orderBy(~mdl_id, glb_models_df[, c("id",
grep(glb_model_metric, names(stats_df), value=TRUE))]))
tmp2_models_df <- glb_models_df[, c("id", setdiff(names(glb_models_df),
grep(glb_model_metric, names(stats_df), value=TRUE)))]
tmp3_models_df <- merge(tmp2_models_df, stats_mrg_df, all.x=TRUE, sort=FALSE)
print(tmp3_models_df)
print(names(tmp3_models_df))
print(glb_models_df <- subset(tmp3_models_df, select=-mdl_id.1))
}
plt_models_df <- glb_models_df[, -grep("SD|Upper|Lower", names(glb_models_df))]
for (var in grep("^min.", names(plt_models_df), value=TRUE)) {
plt_models_df[, sub("min.", "inv.", var)] <-
#ifelse(all(is.na(tmp <- plt_models_df[, var])), NA, 1.0 / tmp)
1.0 / plt_models_df[, var]
plt_models_df <- plt_models_df[ , -grep(var, names(plt_models_df))]
}
print(plt_models_df)## id
## MFO.myMFO_classfr MFO.myMFO_classfr
## Random.myrandom_classfr Random.myrandom_classfr
## Max.cor.Y.rcv.1X1.glmnet Max.cor.Y.rcv.1X1.glmnet
## Max.cor.Y.rcv.3X1.glmnet Max.cor.Y.rcv.3X1.glmnet
## Max.cor.Y.rcv.3X3.glmnet Max.cor.Y.rcv.3X3.glmnet
## Max.cor.Y.rcv.3X5.glmnet Max.cor.Y.rcv.3X5.glmnet
## Max.cor.Y.rcv.5X1.glmnet Max.cor.Y.rcv.5X1.glmnet
## Max.cor.Y.rcv.5X3.glmnet Max.cor.Y.rcv.5X3.glmnet
## Max.cor.Y.rcv.5X5.glmnet Max.cor.Y.rcv.5X5.glmnet
## Max.cor.Y.rcv.1X1.cp.0.rpart Max.cor.Y.rcv.1X1.cp.0.rpart
## Max.cor.Y.rpart Max.cor.Y.rpart
## Interact.High.cor.Y.glmnet Interact.High.cor.Y.glmnet
## Low.cor.X.glmnet Low.cor.X.glmnet
## RFE.X.glm RFE.X.glm
## RFE.X.bayesglm RFE.X.bayesglm
## RFE.X.glmnet RFE.X.glmnet
## RFE.X.rpart RFE.X.rpart
## RFE.X.gbm RFE.X.gbm
## RFE.X.rf RFE.X.rf
## All.X.glmnet All.X.glmnet
## Max.cor.Y.rcv.3X1.Interact.glmnet Max.cor.Y.rcv.3X1.Interact.glmnet
## feats
## MFO.myMFO_classfr .rnorm
## Random.myrandom_classfr .rnorm
## Max.cor.Y.rcv.1X1.glmnet biddable,startprice
## Max.cor.Y.rcv.3X1.glmnet biddable,startprice
## Max.cor.Y.rcv.3X3.glmnet biddable,startprice
## Max.cor.Y.rcv.3X5.glmnet biddable,startprice
## Max.cor.Y.rcv.5X1.glmnet biddable,startprice
## Max.cor.Y.rcv.5X3.glmnet biddable,startprice
## Max.cor.Y.rcv.5X5.glmnet biddable,startprice
## Max.cor.Y.rcv.1X1.cp.0.rpart biddable,startprice
## Max.cor.Y.rpart biddable,startprice
## Interact.High.cor.Y.glmnet biddable,startprice,biddable:cellular.fctr
## Low.cor.X.glmnet biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice
## RFE.X.glm startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.bayesglm startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.glmnet startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.rpart startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.gbm startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## RFE.X.rf startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,color.fctr,cellular.fctr:carrier.fctr
## All.X.glmnet biddable,.rnorm,storage.fctr,color.fctr,prdl.descr.my.fctr,cellular.fctr,condition.fctr,startprice,cellular.fctr:carrier.fctr
## Max.cor.Y.rcv.3X1.Interact.glmnet startprice,biddable*
## max.nTuningRuns max.auc.fit
## MFO.myMFO_classfr 0 0.5000000
## Random.myrandom_classfr 0 0.4950888
## Max.cor.Y.rcv.1X1.glmnet 0 0.8570235
## Max.cor.Y.rcv.3X1.glmnet 25 0.8560789
## Max.cor.Y.rcv.3X3.glmnet 25 0.8568573
## Max.cor.Y.rcv.3X5.glmnet 25 0.8568573
## Max.cor.Y.rcv.5X1.glmnet 25 0.8560789
## Max.cor.Y.rcv.5X3.glmnet 25 0.8568573
## Max.cor.Y.rcv.5X5.glmnet 25 0.8568573
## Max.cor.Y.rcv.1X1.cp.0.rpart 0 0.8939014
## Max.cor.Y.rpart 5 0.8162184
## Interact.High.cor.Y.glmnet 25 0.8567917
## Low.cor.X.glmnet 25 0.8903306
## RFE.X.glm 1 0.8930683
## RFE.X.bayesglm 1 0.8919881
## RFE.X.glmnet 25 0.8767777
## RFE.X.rpart 5 0.8184969
## RFE.X.gbm 25 0.9409210
## RFE.X.rf 5 0.9999978
## All.X.glmnet 25 0.8767777
## Max.cor.Y.rcv.3X1.Interact.glmnet 25 0.8568748
## opt.prob.threshold.fit max.f.score.fit
## MFO.myMFO_classfr 0.5 0.0000000
## Random.myrandom_classfr 0.4 0.6329294
## Max.cor.Y.rcv.1X1.glmnet 0.4 0.7714286
## Max.cor.Y.rcv.3X1.glmnet 0.4 0.7600849
## Max.cor.Y.rcv.3X3.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.3X5.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.5X1.glmnet 0.4 0.7600849
## Max.cor.Y.rcv.5X3.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.5X5.glmnet 0.4 0.7647691
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.4 0.8125677
## Max.cor.Y.rpart 0.6 0.7541401
## Interact.High.cor.Y.glmnet 0.4 0.7617021
## Low.cor.X.glmnet 0.5 0.8013937
## RFE.X.glm 0.6 0.8044010
## RFE.X.bayesglm 0.5 0.7972028
## RFE.X.glmnet 0.5 0.7792208
## RFE.X.rpart 0.7 0.7658473
## RFE.X.gbm 0.4 0.8532423
## RFE.X.rf 0.5 0.9988726
## All.X.glmnet 0.5 0.7792208
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.4 0.7636364
## max.Accuracy.fit max.Kappa.fit .fit
## MFO.myMFO_classfr 0.5370177 0.0000000 0.5370177
## Random.myrandom_classfr 0.4629823 0.0000000 0.4629823
## Max.cor.Y.rcv.1X1.glmnet 0.7831074 0.5653087 0.7831074
## Max.cor.Y.rcv.3X1.glmnet 0.7883033 0.5717602 0.7643379
## Max.cor.Y.rcv.3X3.glmnet 0.7876154 0.5701499 0.7716371
## Max.cor.Y.rcv.3X5.glmnet 0.7879043 0.5707133 0.7716371
## Max.cor.Y.rcv.5X1.glmnet 0.7872982 0.5698703 0.7643379
## Max.cor.Y.rcv.5X3.glmnet 0.7879781 0.5707928 0.7716371
## Max.cor.Y.rcv.5X5.glmnet 0.7874989 0.5698279 0.7716371
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.8196038 0.6391797 0.8196038
## Max.cor.Y.rpart 0.7824049 0.5580889 0.7987487
## Interact.High.cor.Y.glmnet 0.7872682 0.5695577 0.7664234
## Low.cor.X.glmnet 0.7942268 0.5846679 0.8216893
## RFE.X.glm 0.7900546 0.5764108 0.8331595
## RFE.X.bayesglm 0.7904040 0.5769613 0.8185610
## RFE.X.glmnet 0.7949168 0.5850163 0.8050052
## RFE.X.rpart 0.7921358 0.5756141 0.8112617
## RFE.X.gbm 0.8057003 0.6066585 0.8654849
## RFE.X.rf 0.8216893 0.6393413 0.9989572
## All.X.glmnet 0.7949168 0.5850163 0.8050052
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7872682 0.5694675 0.7695516
## max.auc.OOB opt.prob.threshold.OOB
## MFO.myMFO_classfr 0.5000000 0.5
## Random.myrandom_classfr 0.5162111 0.4
## Max.cor.Y.rcv.1X1.glmnet 0.8253362 0.5
## Max.cor.Y.rcv.3X1.glmnet 0.8266524 0.5
## Max.cor.Y.rcv.3X3.glmnet 0.8260961 0.5
## Max.cor.Y.rcv.3X5.glmnet 0.8260961 0.5
## Max.cor.Y.rcv.5X1.glmnet 0.8266524 0.5
## Max.cor.Y.rcv.5X3.glmnet 0.8260961 0.5
## Max.cor.Y.rcv.5X5.glmnet 0.8260961 0.5
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.7971655 0.3
## Max.cor.Y.rpart 0.8095598 0.3
## Interact.High.cor.Y.glmnet 0.8262948 0.5
## Low.cor.X.glmnet 0.8131655 0.5
## RFE.X.glm 0.8055666 0.5
## RFE.X.bayesglm 0.8111689 0.5
## RFE.X.glmnet 0.8178863 0.5
## RFE.X.rpart 0.8092816 0.3
## RFE.X.gbm 0.8223538 0.4
## RFE.X.rf 0.8240549 0.4
## All.X.glmnet 0.8178863 0.5
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.8262948 0.5
## max.f.score.OOB max.Accuracy.OOB
## MFO.myMFO_classfr 0.0000000 0.5377778
## Random.myrandom_classfr 0.6322188 0.4622222
## Max.cor.Y.rcv.1X1.glmnet 0.7474747 0.7777778
## Max.cor.Y.rcv.3X1.glmnet 0.7568922 0.7844444
## Max.cor.Y.rcv.3X3.glmnet 0.7522013 0.7811111
## Max.cor.Y.rcv.3X5.glmnet 0.7522013 0.7811111
## Max.cor.Y.rcv.5X1.glmnet 0.7568922 0.7844444
## Max.cor.Y.rcv.5X3.glmnet 0.7522013 0.7811111
## Max.cor.Y.rcv.5X5.glmnet 0.7522013 0.7811111
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.7260870 0.7200000
## Max.cor.Y.rpart 0.7560976 0.7777778
## Interact.High.cor.Y.glmnet 0.7522013 0.7811111
## Low.cor.X.glmnet 0.7436209 0.7655556
## RFE.X.glm 0.7406514 0.7611111
## RFE.X.bayesglm 0.7442424 0.7655556
## RFE.X.glmnet 0.7573529 0.7800000
## RFE.X.rpart 0.7560976 0.7777778
## RFE.X.gbm 0.7311321 0.7466667
## RFE.X.rf 0.7303754 0.7366667
## All.X.glmnet 0.7573529 0.7800000
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.7522013 0.7811111
## max.Kappa.OOB .OOB
## MFO.myMFO_classfr 0.0000000 0.5377778
## Random.myrandom_classfr 0.0000000 0.4622222
## Max.cor.Y.rcv.1X1.glmnet 0.5499640 0.7777778
## Max.cor.Y.rcv.3X1.glmnet 0.5639099 0.7844444
## Max.cor.Y.rcv.3X3.glmnet 0.5569405 0.7811111
## Max.cor.Y.rcv.3X5.glmnet 0.5569405 0.7811111
## Max.cor.Y.rcv.5X1.glmnet 0.5639099 0.7844444
## Max.cor.Y.rcv.5X3.glmnet 0.5569405 0.7811111
## Max.cor.Y.rcv.5X5.glmnet 0.5569405 0.7811111
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.4450317 0.7200000
## Max.cor.Y.rpart 0.5520962 0.7777778
## Interact.High.cor.Y.glmnet 0.5569405 0.7811111
## Low.cor.X.glmnet 0.5277012 0.7655556
## RFE.X.glm 0.5192355 0.7611111
## RFE.X.bayesglm 0.5278609 0.7655556
## RFE.X.glmnet 0.5562749 0.7800000
## RFE.X.rpart 0.5520962 0.7777778
## RFE.X.gbm 0.4917974 0.7466667
## RFE.X.rf 0.4744804 0.7366667
## All.X.glmnet 0.5562749 0.7800000
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.5569405 0.7811111
## inv.elapsedtime.everything
## MFO.myMFO_classfr 3.7735849
## Random.myrandom_classfr 3.9525692
## Max.cor.Y.rcv.1X1.glmnet 1.5384615
## Max.cor.Y.rcv.3X1.glmnet 0.7315289
## Max.cor.Y.rcv.3X3.glmnet 0.5452563
## Max.cor.Y.rcv.3X5.glmnet 0.4282655
## Max.cor.Y.rcv.5X1.glmnet 0.6418485
## Max.cor.Y.rcv.5X3.glmnet 0.4440497
## Max.cor.Y.rcv.5X5.glmnet 0.3485535
## Max.cor.Y.rcv.1X1.cp.0.rpart 1.4430014
## Max.cor.Y.rpart 0.6230530
## Interact.High.cor.Y.glmnet 0.4506534
## Low.cor.X.glmnet 0.3852080
## RFE.X.glm 0.5422993
## RFE.X.bayesglm 0.3547357
## RFE.X.glmnet 0.3738318
## RFE.X.rpart 0.5473454
## RFE.X.gbm 0.1415228
## RFE.X.rf 0.1079797
## All.X.glmnet 0.3558719
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.4608295
## inv.elapsedtime.final
## MFO.myMFO_classfr 333.3333333
## Random.myrandom_classfr 500.0000000
## Max.cor.Y.rcv.1X1.glmnet 50.0000000
## Max.cor.Y.rcv.3X1.glmnet 58.8235294
## Max.cor.Y.rcv.3X3.glmnet 62.5000000
## Max.cor.Y.rcv.3X5.glmnet 58.8235294
## Max.cor.Y.rcv.5X1.glmnet 71.4285714
## Max.cor.Y.rcv.5X3.glmnet 58.8235294
## Max.cor.Y.rcv.5X5.glmnet 62.5000000
## Max.cor.Y.rcv.1X1.cp.0.rpart 83.3333333
## Max.cor.Y.rpart 83.3333333
## Interact.High.cor.Y.glmnet 62.5000000
## Low.cor.X.glmnet 12.0481928
## RFE.X.glm 11.4942529
## RFE.X.bayesglm 8.0000000
## RFE.X.glmnet 10.4166667
## RFE.X.rpart 21.7391304
## RFE.X.gbm 1.9120459
## RFE.X.rf 0.2715915
## All.X.glmnet 10.5263158
## Max.cor.Y.rcv.3X1.Interact.glmnet 52.6315789print(myplot_radar(radar_inp_df=plt_models_df))## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Set1 is 9
## Returning the palette you asked for with that many colors## Warning: The shape palette can deal with a maximum of 6 discrete values
## because more than 6 becomes difficult to discriminate; you have
## 21. Consider specifying shapes manually if you must have them.## Warning: Removed 240 rows containing missing values (geom_point).## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Set1 is 9
## Returning the palette you asked for with that many colors## Warning: The shape palette can deal with a maximum of 6 discrete values
## because more than 6 becomes difficult to discriminate; you have
## 21. Consider specifying shapes manually if you must have them.
# print(myplot_radar(radar_inp_df=subset(plt_models_df,
# !(mdl_id %in% grep("random|MFO", plt_models_df$id, value=TRUE)))))
# Compute CI for <metric>SD
glb_models_df <- mutate(glb_models_df,
max.df = ifelse(max.nTuningRuns > 1, max.nTuningRuns - 1, NA),
min.sd2ci.scaler = ifelse(is.na(max.df), NA, qt(0.975, max.df)))
for (var in grep("SD", names(glb_models_df), value=TRUE)) {
# Does CI alredy exist ?
var_components <- unlist(strsplit(var, "SD"))
varActul <- paste0(var_components[1], var_components[2])
varUpper <- paste0(var_components[1], "Upper", var_components[2])
varLower <- paste0(var_components[1], "Lower", var_components[2])
if (varUpper %in% names(glb_models_df)) {
warning(varUpper, " already exists in glb_models_df")
# Assuming Lower also exists
next
}
print(sprintf("var:%s", var))
# CI is dependent on sample size in t distribution; df=n-1
glb_models_df[, varUpper] <- glb_models_df[, varActul] +
glb_models_df[, "min.sd2ci.scaler"] * glb_models_df[, var]
glb_models_df[, varLower] <- glb_models_df[, varActul] -
glb_models_df[, "min.sd2ci.scaler"] * glb_models_df[, var]
}## Warning: max.AccuracyUpper.fit already exists in glb_models_df## [1] "var:max.KappaSD.fit"# Plot metrics with CI
plt_models_df <- glb_models_df[, "id", FALSE]
pltCI_models_df <- glb_models_df[, "id", FALSE]
for (var in grep("Upper", names(glb_models_df), value=TRUE)) {
var_components <- unlist(strsplit(var, "Upper"))
col_name <- unlist(paste(var_components, collapse=""))
plt_models_df[, col_name] <- glb_models_df[, col_name]
for (name in paste0(var_components[1], c("Upper", "Lower"), var_components[2]))
pltCI_models_df[, name] <- glb_models_df[, name]
}
build_statsCI_data <- function(plt_models_df) {
mltd_models_df <- melt(plt_models_df, id.vars="id")
mltd_models_df$data <- sapply(1:nrow(mltd_models_df),
function(row_ix) tail(unlist(strsplit(as.character(
mltd_models_df[row_ix, "variable"]), "[.]")), 1))
mltd_models_df$label <- sapply(1:nrow(mltd_models_df),
function(row_ix) head(unlist(strsplit(as.character(
mltd_models_df[row_ix, "variable"]),
paste0(".", mltd_models_df[row_ix, "data"]))), 1))
#print(mltd_models_df)
return(mltd_models_df)
}
mltd_models_df <- build_statsCI_data(plt_models_df)
mltdCI_models_df <- melt(pltCI_models_df, id.vars="id")
for (row_ix in 1:nrow(mltdCI_models_df)) {
for (type in c("Upper", "Lower")) {
if (length(var_components <- unlist(strsplit(
as.character(mltdCI_models_df[row_ix, "variable"]), type))) > 1) {
#print(sprintf("row_ix:%d; type:%s; ", row_ix, type))
mltdCI_models_df[row_ix, "label"] <- var_components[1]
mltdCI_models_df[row_ix, "data"] <-
unlist(strsplit(var_components[2], "[.]"))[2]
mltdCI_models_df[row_ix, "type"] <- type
break
}
}
}
wideCI_models_df <- reshape(subset(mltdCI_models_df, select=-variable),
timevar="type",
idvar=setdiff(names(mltdCI_models_df), c("type", "value", "variable")),
direction="wide")
#print(wideCI_models_df)
mrgdCI_models_df <- merge(wideCI_models_df, mltd_models_df, all.x=TRUE)
#print(mrgdCI_models_df)
# Merge stats back in if CIs don't exist
goback_vars <- c()
for (var in unique(mltd_models_df$label)) {
for (type in unique(mltd_models_df$data)) {
var_type <- paste0(var, ".", type)
# if this data is already present, next
if (var_type %in% unique(paste(mltd_models_df$label, mltd_models_df$data,
sep=".")))
next
#print(sprintf("var_type:%s", var_type))
goback_vars <- c(goback_vars, var_type)
}
}
if (length(goback_vars) > 0) {
mltd_goback_df <- build_statsCI_data(glb_models_df[, c("id", goback_vars)])
mltd_models_df <- rbind(mltd_models_df, mltd_goback_df)
}
# mltd_models_df <- merge(mltd_models_df, glb_models_df[, c("id", "model_method")],
# all.x=TRUE)
png(paste0(glb_out_pfx, "models_bar.png"), width=480*3, height=480*2)
#print(gp <- myplot_bar(mltd_models_df, "id", "value", colorcol_name="model_method") +
print(gp <- myplot_bar(df=mltd_models_df, xcol_name="id", ycol_names="value") +
geom_errorbar(data=mrgdCI_models_df,
mapping=aes(x=mdl_id, ymax=value.Upper, ymin=value.Lower), width=0.5) +
facet_grid(label ~ data, scales="free") +
theme(axis.text.x = element_text(angle = 90,vjust = 0.5)))## Warning: Removed 7 rows containing missing values (geom_errorbar).dev.off()## quartz_off_screen
## 2print(gp)## Warning: Removed 7 rows containing missing values (geom_errorbar).
dsp_models_cols <- c("id",
glb_model_evl_criteria[glb_model_evl_criteria %in% names(glb_models_df)])
if (glb_is_classification && glb_is_binomial)
dsp_models_cols <- c(dsp_models_cols, "opt.prob.threshold.OOB")
print(dsp_models_df <- orderBy(get_model_sel_frmla(), glb_models_df)[, dsp_models_cols])## id max.auc.OOB max.Accuracy.OOB
## 4 Max.cor.Y.rcv.3X1.glmnet 0.8266524 0.7844444
## 7 Max.cor.Y.rcv.5X1.glmnet 0.8266524 0.7844444
## 12 Interact.High.cor.Y.glmnet 0.8262948 0.7811111
## 21 Max.cor.Y.rcv.3X1.Interact.glmnet 0.8262948 0.7811111
## 5 Max.cor.Y.rcv.3X3.glmnet 0.8260961 0.7811111
## 6 Max.cor.Y.rcv.3X5.glmnet 0.8260961 0.7811111
## 8 Max.cor.Y.rcv.5X3.glmnet 0.8260961 0.7811111
## 9 Max.cor.Y.rcv.5X5.glmnet 0.8260961 0.7811111
## 3 Max.cor.Y.rcv.1X1.glmnet 0.8253362 0.7777778
## 19 RFE.X.rf 0.8240549 0.7366667
## 18 RFE.X.gbm 0.8223538 0.7466667
## 16 RFE.X.glmnet 0.8178863 0.7800000
## 20 All.X.glmnet 0.8178863 0.7800000
## 13 Low.cor.X.glmnet 0.8131655 0.7655556
## 15 RFE.X.bayesglm 0.8111689 0.7655556
## 11 Max.cor.Y.rpart 0.8095598 0.7777778
## 17 RFE.X.rpart 0.8092816 0.7777778
## 14 RFE.X.glm 0.8055666 0.7611111
## 10 Max.cor.Y.rcv.1X1.cp.0.rpart 0.7971655 0.7200000
## 2 Random.myrandom_classfr 0.5162111 0.4622222
## 1 MFO.myMFO_classfr 0.5000000 0.5377778
## max.Kappa.OOB opt.prob.threshold.OOB
## 4 0.5639099 0.5
## 7 0.5639099 0.5
## 12 0.5569405 0.5
## 21 0.5569405 0.5
## 5 0.5569405 0.5
## 6 0.5569405 0.5
## 8 0.5569405 0.5
## 9 0.5569405 0.5
## 3 0.5499640 0.5
## 19 0.4744804 0.4
## 18 0.4917974 0.4
## 16 0.5562749 0.5
## 20 0.5562749 0.5
## 13 0.5277012 0.5
## 15 0.5278609 0.5
## 11 0.5520962 0.3
## 17 0.5520962 0.3
## 14 0.5192355 0.5
## 10 0.4450317 0.3
## 2 0.0000000 0.4
## 1 0.0000000 0.5print(myplot_radar(radar_inp_df = dsp_models_df))## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Set1 is 9
## Returning the palette you asked for with that many colors## Warning: The shape palette can deal with a maximum of 6 discrete values
## because more than 6 becomes difficult to discriminate; you have
## 21. Consider specifying shapes manually if you must have them.## Warning: Removed 75 rows containing missing values (geom_point).## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Set1 is 9
## Returning the palette you asked for with that many colors## Warning: The shape palette can deal with a maximum of 6 discrete values
## because more than 6 becomes difficult to discriminate; you have
## 21. Consider specifying shapes manually if you must have them.
print("Metrics used for model selection:"); print(get_model_sel_frmla())## [1] "Metrics used for model selection:"## ~-max.auc.OOB - max.Accuracy.OOB - max.Kappa.OOB - opt.prob.threshold.OOB
## <environment: 0x7faed6fbacd8>print(sprintf("Best model id: %s", dsp_models_df[1, "id"]))## [1] "Best model id: Max.cor.Y.rcv.3X1.glmnet"glb_get_predictions <- function(df, mdl_id, rsp_var_out, prob_threshold_def=NULL, verbose=FALSE) {
mdl <- glb_models_lst[[mdl_id]]
#rsp_var_out <- paste0(rsp_var_out, mdl_id)
rsp_var_out <- paste0(glb_rsp_var, ".predict.")
predct_var_name <- paste0(rsp_var_out, mdl_id)
predct_prob_var_name <- paste0(rsp_var_out, mdl_id, ".prob")
predct_accurate_var_name <- paste0(rsp_var_out, mdl_id, ".accurate")
predct_error_var_name <- paste0(rsp_var_out, mdl_id, ".err")
predct_erabs_var_name <- paste0(rsp_var_out, mdl_id, ".err.abs")
if (glb_is_regression) {
df[, predct_var_name] <- predict(mdl, newdata=df, type="raw")
if (verbose) print(myplot_scatter(df, glb_rsp_var, predct_var_name) +
facet_wrap(reformulate(glb_category_var), scales = "free") +
stat_smooth(method="glm"))
df[, predct_error_var_name] <- df[, predct_var_name] - df[, glb_rsp_var]
if (verbose) print(myplot_scatter(df, predct_var_name, predct_error_var_name) +
#facet_wrap(reformulate(glb_category_var), scales = "free") +
stat_smooth(method="auto"))
if (verbose) print(myplot_scatter(df, glb_rsp_var, predct_error_var_name) +
#facet_wrap(reformulate(glb_category_var), scales = "free") +
stat_smooth(method="glm"))
df[, predct_erabs_var_name] <- abs(df[, predct_error_var_name])
if (verbose) print(head(orderBy(reformulate(c("-", predct_erabs_var_name)), df)))
df[, predct_accurate_var_name] <- (df[, glb_rsp_var] == df[, predct_var_name])
}
if (glb_is_classification && glb_is_binomial) {
prob_threshold <- glb_models_df[glb_models_df$id == mdl_id,
"opt.prob.threshold.OOB"]
if (is.null(prob_threshold) || is.na(prob_threshold)) {
warning("Using default probability threshold: ", prob_threshold_def)
if (is.null(prob_threshold <- prob_threshold_def))
stop("Default probability threshold is NULL")
}
df[, predct_prob_var_name] <- predict(mdl, newdata = df, type = "prob")[, 2]
df[, predct_var_name] <-
factor(levels(df[, glb_rsp_var])[
(df[, predct_prob_var_name] >=
prob_threshold) * 1 + 1], levels(df[, glb_rsp_var]))
# if (verbose) print(myplot_scatter(df, glb_rsp_var, predct_var_name) +
# facet_wrap(reformulate(glb_category_var), scales = "free") +
# stat_smooth(method="glm"))
df[, predct_error_var_name] <- df[, predct_var_name] == df[, glb_rsp_var]
# if (verbose) print(myplot_scatter(df, predct_var_name, predct_error_var_name) +
# #facet_wrap(reformulate(glb_category_var), scales = "free") +
# stat_smooth(method="auto"))
# if (verbose) print(myplot_scatter(df, glb_rsp_var, predct_error_var_name) +
# #facet_wrap(reformulate(glb_category_var), scales = "free") +
# stat_smooth(method="glm"))
# if prediction is a TP (true +ve), measure distance from 1.0
tp <- which((df[, predct_var_name] == df[, glb_rsp_var]) &
(df[, predct_var_name] == levels(df[, glb_rsp_var])[2]))
df[tp, predct_erabs_var_name] <- abs(1 - df[tp, predct_prob_var_name])
#rowIx <- which.max(df[tp, predct_erabs_var_name]); df[tp, c(glb_id_var, glb_rsp_var, predct_var_name, predct_prob_var_name, predct_erabs_var_name)][rowIx, ]
# if prediction is a TN (true -ve), measure distance from 0.0
tn <- which((df[, predct_var_name] == df[, glb_rsp_var]) &
(df[, predct_var_name] == levels(df[, glb_rsp_var])[1]))
df[tn, predct_erabs_var_name] <- abs(0 - df[tn, predct_prob_var_name])
#rowIx <- which.max(df[tn, predct_erabs_var_name]); df[tn, c(glb_id_var, glb_rsp_var, predct_var_name, predct_prob_var_name, predct_erabs_var_name)][rowIx, ]
# if prediction is a FP (flse +ve), measure distance from 0.0
fp <- which((df[, predct_var_name] != df[, glb_rsp_var]) &
(df[, predct_var_name] == levels(df[, glb_rsp_var])[2]))
df[fp, predct_erabs_var_name] <- abs(0 - df[fp, predct_prob_var_name])
#rowIx <- which.max(df[fp, predct_erabs_var_name]); df[fp, c(glb_id_var, glb_rsp_var, predct_var_name, predct_prob_var_name, predct_erabs_var_name)][rowIx, ]
# if prediction is a FN (flse -ve), measure distance from 1.0
fn <- which((df[, predct_var_name] != df[, glb_rsp_var]) &
(df[, predct_var_name] == levels(df[, glb_rsp_var])[1]))
df[fn, predct_erabs_var_name] <- abs(1 - df[fn, predct_prob_var_name])
#rowIx <- which.max(df[fn, predct_erabs_var_name]); df[fn, c(glb_id_var, glb_rsp_var, predct_var_name, predct_prob_var_name, predct_erabs_var_name)][rowIx, ]
if (verbose) print(head(orderBy(reformulate(c("-", predct_erabs_var_name)), df)))
df[, predct_accurate_var_name] <- (df[, glb_rsp_var] == df[, predct_var_name])
}
if (glb_is_classification && !glb_is_binomial) {
df[, predct_var_name] <- predict(mdl, newdata = df, type = "raw")
df[, paste0(predct_var_name, ".prob")] <-
predict(mdl, newdata = df, type = "prob")
stop("Multinomial prediction error calculation needs to be implemented...")
}
return(df)
}
#stop(here"); glb_to_sav(); glb_allobs_df <- sav_allobs_df; glb_trnobs_df <- sav_trnobs_df; glb_fitobs_df <- sav_fitobs_df; glb_OOBobs_df <- sav_OOBobs_df; sav_models_df <- glb_models_df; glb_models_df <- sav_models_df; glb_featsimp_df <- sav_featsimp_df
myget_category_stats <- function(obs_df, mdl_id, label) {
require(dplyr)
require(lazyeval)
predct_var_name <- paste0(glb_rsp_var_out, mdl_id)
predct_error_var_name <- paste0(glb_rsp_var_out, mdl_id, ".err.abs")
if (!predct_var_name %in% names(obs_df))
obs_df <- glb_get_predictions(obs_df, mdl_id, glb_rsp_var_out)
tmp_obs_df <- obs_df %>%
dplyr::select_(glb_category_var, glb_rsp_var, predct_var_name, predct_error_var_name)
#dplyr::rename(startprice.log10.predict.RFE.X.glmnet.err=error_abs_OOB)
names(tmp_obs_df)[length(names(tmp_obs_df))] <- paste0("err.abs.", label)
ret_ctgry_df <- tmp_obs_df %>%
dplyr::group_by_(glb_category_var) %>%
dplyr::summarise_(#interp(~sum(abs(var)), var=as.name(glb_rsp_var)),
interp(~sum(var), var=as.name(paste0("err.abs.", label))),
interp(~mean(var), var=as.name(paste0("err.abs.", label))),
interp(~n()))
names(ret_ctgry_df) <- c(glb_category_var,
#paste0(glb_rsp_var, ".abs.", label, ".sum"),
paste0("err.abs.", label, ".sum"),
paste0("err.abs.", label, ".mean"),
paste0(".n.", label))
ret_ctgry_df <- dplyr::ungroup(ret_ctgry_df)
#colSums(ret_ctgry_df[, -grep(glb_category_var, names(ret_ctgry_df))])
return(ret_ctgry_df)
}
#print(colSums((ctgry_df <- myget_category_stats(obs_df=glb_fitobs_df, mdl_id="", label="fit"))[, -grep(glb_category_var, names(ctgry_df))]))
if (!is.null(glb_mdl_ensemble)) {
mdl_id_pfx <- "Ensemble"
if (#(glb_is_regression) |
((glb_is_classification) & (!glb_is_binomial)))
stop("Ensemble models not implemented yet for multinomial classification")
if (glb_mdl_ensemble == "auto") {
mdl_id_pfx <- paste0(mdl_id_pfx, ".auto")
tmp_models_df <- orderBy(get_model_sel_frmla(), glb_models_df)
row.names(tmp_models_df) <- tmp_models_df$id
# mdl_threshold_pos <- min(which(tmp_models_df$id %in%
# c("MFO.myMFO_classfr", "Baseline.mybaseln_classfr"))) - 1
mdl_threshold_pos <-
min(which(grepl("MFO|Random|Baseline", tmp_models_df$id))) - 1
glb_mdl_ensemble <- tmp_models_df$id[1:mdl_threshold_pos]
}
for (mdl_id in glb_mdl_ensemble) {
glb_fitobs_df <- glb_get_predictions(df=glb_fitobs_df, mdl_id, glb_rsp_var_out)
glb_OOBobs_df <- glb_get_predictions(df=glb_OOBobs_df, mdl_id, glb_rsp_var_out)
}
#mdl_id_pfx <- "Ensemble.RFE"; mdlId <- paste0(mdl_id_pfx, ".glmnet")
#glb_mdl_ensemble <- gsub(glb_rsp_var_out, "", grep("RFE\\.X\\.(?!Interact)", row.names(glb_featsimp_df), perl = TRUE, value = TRUE), fixed = TRUE)
#varImp(glb_models_lst[[mdlId]])
#cor_df <- data.frame(cor=cor(glb_fitobs_df[, glb_rsp_var], glb_fitobs_df[, paste(glb_rsp_var_out, glb_mdl_ensemble)], use="pairwise.complete.obs"))
#glb_fitobs_df <- glb_get_predictions(df=glb_fitobs_df, "Ensemble.glmnet", glb_rsp_var_out);print(colSums((ctgry_df <- myget_category_stats(obs_df=glb_fitobs_df, mdl_id="Ensemble.glmnet", label="fit"))[, -grep(glb_category_var, names(ctgry_df))]))
### bid0_sp
# Better than MFO; models.n=28; min.RMSE.fit=0.0521233; err.abs.fit.sum=7.3631895
# old: Top x from auto; models.n= 5; min.RMSE.fit=0.06311047; err.abs.fit.sum=9.5937080
# RFE only ; models.n=16; min.RMSE.fit=0.05148588; err.abs.fit.sum=7.2875091
# RFE subset only ;models.n= 5; min.RMSE.fit=0.06040702; err.abs.fit.sum=9.059088
# RFE subset only ;models.n= 9; min.RMSE.fit=0.05933167; err.abs.fit.sum=8.7421288
# RFE subset only ;models.n=15; min.RMSE.fit=0.0584607; err.abs.fit.sum=8.5902066
# RFE subset only ;models.n=17; min.RMSE.fit=0.05496899; err.abs.fit.sum=8.0170431
# RFE subset only ;models.n=18; min.RMSE.fit=0.05441577; err.abs.fit.sum=7.837223
# RFE subset only ;models.n=16; min.RMSE.fit=0.05441577; err.abs.fit.sum=7.837223
### bid0_sp
### bid1_sp
# "auto"; err.abs.fit.sum=76.699774; min.RMSE.fit=0.2186429
# "RFE.X.*"; err.abs.fit.sum=; min.RMSE.fit=0.221114
### bid1_sp
indep_vars <- paste(glb_rsp_var_out, glb_mdl_ensemble, sep = "")
if (glb_is_classification)
indep_vars <- paste(indep_vars, ".prob", sep = "")
# indep_vars <- grep(glb_rsp_var_out, names(glb_fitobs_df), fixed=TRUE, value=TRUE)
# if (glb_is_regression)
# indep_vars <- indep_vars[!grepl("(err\\.abs|accurate)$", indep_vars)]
# if (glb_is_classification && glb_is_binomial)
# indep_vars <- grep("prob$", indep_vars, value=TRUE) else
# indep_vars <- indep_vars[!grepl("err$", indep_vars)]
#rfe_fit_ens_results <- myrun_rfe(glb_fitobs_df, indep_vars)
for (method in c("glmnet")) {
#sav_models_df <- glb_models_df; all.equal(sav_models_df, glb_models_df)
#glb_models_df <- sav_models_df; print(glb_models_df$id)
ret_lst <- myfit_mdl(mdl_specs_lst=myinit_mdl_specs_lst(mdl_specs_lst=list(
id.prefix=mdl_id_pfx,
type="regression", tune.df=NULL,
trainControl.method="repeatedcv",
trainControl.number=glb_rcv_n_folds, trainControl.repeats=glb_rcv_n_repeats,
train.method=method)),
indep_vars=indep_vars, rsp_var=glb_rsp_var,
fit_df=glb_fitobs_df, OOB_df=glb_OOBobs_df)
}
dsp_models_df <- get_dsp_models_df()
}
if (is.null(glb_sel_mdl_id))
glb_sel_mdl_id <- dsp_models_df[1, "id"] else
print(sprintf("User specified selection: %s", glb_sel_mdl_id)) ## [1] "User specified selection: RFE.X.glmnet"myprint_mdl(glb_sel_mdl <- glb_models_lst[[glb_sel_mdl_id]])
## Length Class Mode
## a0 86 -none- numeric
## beta 4988 dgCMatrix S4
## df 86 -none- numeric
## dim 2 -none- numeric
## lambda 86 -none- numeric
## dev.ratio 86 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 58 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept)
## 0.212280420
## biddable
## 1.517695590
## cellular.fctr1
## 0.111291692
## cellular.fctrUnknown
## -0.240821550
## condition.fctrSeller refurbished
## -0.303613584
## prdl.descr.my.fctrUnknown#1
## -0.001428038
## prdl.descr.my.fctriPad2#0
## 0.578460972
## prdl.descr.my.fctriPad2#1
## -0.138431499
## prdl.descr.my.fctriPad3#0
## 0.193090727
## prdl.descr.my.fctriPad4#0
## 0.239392597
## prdl.descr.my.fctriPad4#1
## -0.926942426
## prdl.descr.my.fctriPadAir#0
## 0.356239530
## prdl.descr.my.fctriPadAir2#0
## 0.496926272
## prdl.descr.my.fctriPadAir2#1
## 0.329623264
## prdl.descr.my.fctriPadmini#0
## -0.054182251
## prdl.descr.my.fctriPadmini2#0
## 0.071308286
## prdl.descr.my.fctriPadmini3#0
## -0.058991667
## prdl.descr.my.fctriPadminiRetina#0
## 0.615538025
## startprice
## -0.005378116
## storage.fctr32
## -0.012610177
## storage.fctr64
## 0.062650081
## cellular.fctr1:carrier.fctrT-Mobile
## 0.198632683
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.232792402
## cellular.fctr1:carrier.fctrVerizon
## 0.039311991
## [1] "max lambda < lambdaOpt:"
## (Intercept)
## 0.242008407
## biddable
## 1.519615706
## cellular.fctr1
## 0.118581035
## cellular.fctrUnknown
## -0.251456870
## condition.fctrSeller refurbished
## -0.341299917
## prdl.descr.my.fctrUnknown#1
## -0.016706048
## prdl.descr.my.fctriPad2#0
## 0.620952079
## prdl.descr.my.fctriPad2#1
## -0.167333182
## prdl.descr.my.fctriPad3#0
## 0.225106750
## prdl.descr.my.fctriPad4#0
## 0.288944274
## prdl.descr.my.fctriPad4#1
## -0.963264406
## prdl.descr.my.fctriPadAir#0
## 0.426395470
## prdl.descr.my.fctriPadAir2#0
## 0.587153313
## prdl.descr.my.fctriPadAir2#1
## 0.421222039
## prdl.descr.my.fctriPadmini#0
## -0.066928131
## prdl.descr.my.fctriPadmini2#0
## 0.133141513
## prdl.descr.my.fctriPadmini3#0
## -0.059558089
## prdl.descr.my.fctriPadminiRetina#0
## 0.775062295
## startprice
## -0.005612763
## storage.fctr32
## -0.029076706
## storage.fctr64
## 0.078190707
## cellular.fctr1:carrier.fctrT-Mobile
## 0.294925866
## cellular.fctr1:carrier.fctrUnknown
## 0.024140616
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.241582561
## cellular.fctr1:carrier.fctrVerizon
## 0.069556614## [1] TRUE#stop(here"); glb_to_sav()
# From here to save(), this should all be in one function
# these are executed in the same seq twice more:
# fit.data.training & predict.data.new chunks
print(sprintf("%s fit prediction diagnostics:", glb_sel_mdl_id))## [1] "RFE.X.glmnet fit prediction diagnostics:"glb_fitobs_df <- glb_get_predictions(df=glb_fitobs_df, mdl_id=glb_sel_mdl_id,
rsp_var_out=glb_rsp_var_out)
print(sprintf("%s OOB prediction diagnostics:", glb_sel_mdl_id))## [1] "RFE.X.glmnet OOB prediction diagnostics:"glb_OOBobs_df <- glb_get_predictions(df = glb_OOBobs_df, mdl_id = glb_sel_mdl_id,
rsp_var_out = glb_rsp_var_out)
glb_featsimp_df <-
myget_feats_importance(mdl=glb_sel_mdl, featsimp_df=NULL)
glb_featsimp_df[, paste0(glb_sel_mdl_id, ".importance")] <- glb_featsimp_df$importance
#mdl_id <-"RFE.X.glmnet"; glb_featsimp_df <- myget_feats_importance(glb_models_lst[[mdl_id]], glb_featsimp_df); glb_featsimp_df[, paste0(mdl_id, ".importance")] <- glb_featsimp_df$importance; print(glb_featsimp_df)
#print(head(sbst_featsimp_df <- subset(glb_featsimp_df, is.na(RFE.X.glmnet.importance) | (abs(RFE.X.YeoJohnson.glmnet.importance - RFE.X.glmnet.importance) > 0.0001), select=-importance)))
#print(orderBy(~ -cor.y.abs, subset(glb_feats_df, id %in% c(row.names(sbst_featsimp_df), "startprice.dcm1.is9", "D.weight.post.stop.sum"))))
print(glb_featsimp_df)## importance
## biddable 100.00000
## prdl.descr.my.fctriPadminiRetina#0 69.85328
## prdl.descr.my.fctriPad2#0 63.75435
## prdl.descr.my.fctriPadAir2#0 62.34763
## prdl.descr.my.fctriPadAir#0 55.88955
## prdl.descr.my.fctriPadAir2#1 55.66092
## cellular.fctr1:carrier.fctrT-Mobile 50.56798
## prdl.descr.my.fctriPad4#0 50.37100
## prdl.descr.my.fctriPad3#0 47.81549
## prdl.descr.my.fctriPadmini2#0 44.08210
## cellular.fctr1 43.54682
## storage.fctr64 41.91171
## cellular.fctr1:carrier.fctrVerizon 41.54999
## cellular.fctr1:carrier.fctrUnknown 39.72591
## .rnorm 38.77602
## color.fctrGold 38.77602
## color.fctrSpace Gray 38.77602
## color.fctrUnknown 38.77602
## color.fctrWhite 38.77602
## condition.fctrFor parts or not working 38.77602
## condition.fctrManufacturer refurbished 38.77602
## condition.fctrNew 38.77602
## condition.fctrNew other (see details) 38.77602
## prdl.descr.my.fctriPad1#0 38.77602
## prdl.descr.my.fctriPad1#1 38.77602
## prdl.descr.my.fctriPad3#1 38.77602
## prdl.descr.my.fctriPad5#0 38.77602
## prdl.descr.my.fctriPadAir#1 38.77602
## prdl.descr.my.fctriPadmini#1 38.77602
## prdl.descr.my.fctriPadmini2#1 38.77602
## prdl.descr.my.fctriPadmini3#1 38.77602
## prdl.descr.my.fctriPadminiRetina#1 38.77602
## storage.fctr16 38.77602
## storage.fctrUnknown 38.77602
## cellular.fctr0:carrier.fctrNone 38.77602
## cellular.fctr1:carrier.fctrNone 38.77602
## cellular.fctrUnknown:carrier.fctrNone 38.77602
## cellular.fctr0:carrier.fctrOther 38.77602
## cellular.fctr1:carrier.fctrOther 38.77602
## cellular.fctrUnknown:carrier.fctrOther 38.77602
## cellular.fctr0:carrier.fctrSprint 38.77602
## cellular.fctr1:carrier.fctrSprint 38.77602
## cellular.fctrUnknown:carrier.fctrSprint 38.77602
## cellular.fctr0:carrier.fctrT-Mobile 38.77602
## cellular.fctrUnknown:carrier.fctrT-Mobile 38.77602
## cellular.fctr0:carrier.fctrUnknown 38.77602
## cellular.fctr0:carrier.fctrVerizon 38.77602
## cellular.fctrUnknown:carrier.fctrVerizon 38.77602
## startprice 38.55010
## prdl.descr.my.fctrUnknown#1 38.11731
## storage.fctr32 37.62001
## prdl.descr.my.fctriPadmini3#0 36.37694
## prdl.descr.my.fctriPadmini#0 36.09147
## prdl.descr.my.fctriPad2#1 32.06133
## cellular.fctrUnknown:carrier.fctrUnknown 29.05086
## cellular.fctrUnknown 28.65476
## condition.fctrSeller refurbished 25.06043
## prdl.descr.my.fctriPad4#1 0.00000
## RFE.X.glmnet.importance
## biddable 100.00000
## prdl.descr.my.fctriPadminiRetina#0 69.85328
## prdl.descr.my.fctriPad2#0 63.75435
## prdl.descr.my.fctriPadAir2#0 62.34763
## prdl.descr.my.fctriPadAir#0 55.88955
## prdl.descr.my.fctriPadAir2#1 55.66092
## cellular.fctr1:carrier.fctrT-Mobile 50.56798
## prdl.descr.my.fctriPad4#0 50.37100
## prdl.descr.my.fctriPad3#0 47.81549
## prdl.descr.my.fctriPadmini2#0 44.08210
## cellular.fctr1 43.54682
## storage.fctr64 41.91171
## cellular.fctr1:carrier.fctrVerizon 41.54999
## cellular.fctr1:carrier.fctrUnknown 39.72591
## .rnorm 38.77602
## color.fctrGold 38.77602
## color.fctrSpace Gray 38.77602
## color.fctrUnknown 38.77602
## color.fctrWhite 38.77602
## condition.fctrFor parts or not working 38.77602
## condition.fctrManufacturer refurbished 38.77602
## condition.fctrNew 38.77602
## condition.fctrNew other (see details) 38.77602
## prdl.descr.my.fctriPad1#0 38.77602
## prdl.descr.my.fctriPad1#1 38.77602
## prdl.descr.my.fctriPad3#1 38.77602
## prdl.descr.my.fctriPad5#0 38.77602
## prdl.descr.my.fctriPadAir#1 38.77602
## prdl.descr.my.fctriPadmini#1 38.77602
## prdl.descr.my.fctriPadmini2#1 38.77602
## prdl.descr.my.fctriPadmini3#1 38.77602
## prdl.descr.my.fctriPadminiRetina#1 38.77602
## storage.fctr16 38.77602
## storage.fctrUnknown 38.77602
## cellular.fctr0:carrier.fctrNone 38.77602
## cellular.fctr1:carrier.fctrNone 38.77602
## cellular.fctrUnknown:carrier.fctrNone 38.77602
## cellular.fctr0:carrier.fctrOther 38.77602
## cellular.fctr1:carrier.fctrOther 38.77602
## cellular.fctrUnknown:carrier.fctrOther 38.77602
## cellular.fctr0:carrier.fctrSprint 38.77602
## cellular.fctr1:carrier.fctrSprint 38.77602
## cellular.fctrUnknown:carrier.fctrSprint 38.77602
## cellular.fctr0:carrier.fctrT-Mobile 38.77602
## cellular.fctrUnknown:carrier.fctrT-Mobile 38.77602
## cellular.fctr0:carrier.fctrUnknown 38.77602
## cellular.fctr0:carrier.fctrVerizon 38.77602
## cellular.fctrUnknown:carrier.fctrVerizon 38.77602
## startprice 38.55010
## prdl.descr.my.fctrUnknown#1 38.11731
## storage.fctr32 37.62001
## prdl.descr.my.fctriPadmini3#0 36.37694
## prdl.descr.my.fctriPadmini#0 36.09147
## prdl.descr.my.fctriPad2#1 32.06133
## cellular.fctrUnknown:carrier.fctrUnknown 29.05086
## cellular.fctrUnknown 28.65476
## condition.fctrSeller refurbished 25.06043
## prdl.descr.my.fctriPad4#1 0.00000# Used again in fit.data.training & predict.data.new chunks
glb_analytics_diag_plots <- function(obs_df, mdl_id, prob_threshold=NULL) {
if (!is.null(featsimp_df <- glb_featsimp_df)) {
featsimp_df$feat <- gsub("`(.*?)`", "\\1", row.names(featsimp_df))
featsimp_df$feat.interact <- gsub("(.*?):(.*)", "\\2", featsimp_df$feat)
featsimp_df$feat <- gsub("(.*?):(.*)", "\\1", featsimp_df$feat)
featsimp_df$feat.interact <- ifelse(featsimp_df$feat.interact == featsimp_df$feat,
NA, featsimp_df$feat.interact)
featsimp_df$feat <- gsub("(.*?)\\.fctr(.*)", "\\1\\.fctr", featsimp_df$feat)
featsimp_df$feat.interact <- gsub("(.*?)\\.fctr(.*)", "\\1\\.fctr", featsimp_df$feat.interact)
featsimp_df <- orderBy(~ -importance.max, summaryBy(importance ~ feat + feat.interact,
data=featsimp_df, FUN=max))
#rex_str=":(.*)"; txt_vctr=tail(featsimp_df$feat); ret_lst <- regexec(rex_str, txt_vctr); ret_lst <- regmatches(txt_vctr, ret_lst); ret_vctr <- sapply(1:length(ret_lst), function(pos_ix) ifelse(length(ret_lst[[pos_ix]]) > 0, ret_lst[[pos_ix]], "")); print(ret_vctr <- ret_vctr[ret_vctr != ""])
featsimp_df <- subset(featsimp_df, !is.na(importance.max))
if (nrow(featsimp_df) > 5) {
warning("Limiting important feature scatter plots to 5 out of ", nrow(featsimp_df))
featsimp_df <- head(featsimp_df, 5)
}
# if (!all(is.na(featsimp_df$feat.interact)))
# stop("not implemented yet")
rsp_var_out <- paste0(glb_rsp_var_out, mdl_id)
for (var in featsimp_df$feat) {
plot_df <- melt(obs_df, id.vars=var,
measure.vars=c(glb_rsp_var, rsp_var_out))
# if (var == "<feat_name>") print(myplot_scatter(plot_df, var, "value",
# facet_colcol_name="variable") +
# geom_vline(xintercept=<divider_val>, linetype="dotted")) else
print(myplot_scatter(plot_df, var, "value", colorcol_name="variable",
facet_colcol_name="variable", jitter=TRUE) +
guides(color=FALSE))
}
}
if (glb_is_regression) {
if (is.null(featsimp_df) || (nrow(featsimp_df) == 0))
warning("No important features in glb_fin_mdl") else
print(myplot_prediction_regression(df=obs_df,
feat_x=ifelse(nrow(featsimp_df) > 1, featsimp_df$feat[2],
".rownames"),
feat_y=featsimp_df$feat[1],
rsp_var=glb_rsp_var, rsp_var_out=rsp_var_out,
id_vars=glb_id_var)
# + facet_wrap(reformulate(featsimp_df$feat[2])) # if [1 or 2] is a factor
# + geom_point(aes_string(color="<col_name>.fctr")) # to color the plot
)
}
if (glb_is_classification) {
if (is.null(featsimp_df) || (nrow(featsimp_df) == 0))
warning("No features in selected model are statistically important")
else print(myplot_prediction_classification(df=obs_df,
feat_x=ifelse(nrow(featsimp_df) > 1, featsimp_df$feat[2],
".rownames"),
feat_y=featsimp_df$feat[1],
rsp_var=glb_rsp_var,
rsp_var_out=rsp_var_out,
id_vars=glb_id_var,
prob_threshold=prob_threshold)
# + geom_hline(yintercept=<divider_val>, linetype = "dotted")
)
}
}
if (glb_is_classification && glb_is_binomial)
glb_analytics_diag_plots(obs_df = glb_OOBobs_df, mdl_id = glb_sel_mdl_id,
prob_threshold = glb_models_df[glb_models_df$id == glb_sel_mdl_id,
"opt.prob.threshold.OOB"]) else
glb_analytics_diag_plots(obs_df=glb_OOBobs_df, mdl_id=glb_sel_mdl_id) ## Warning in glb_analytics_diag_plots(obs_df = glb_OOBobs_df, mdl_id =
## glb_sel_mdl_id, : Limiting important feature scatter plots to 5 out of 9
## [1] "Min/Max Boundaries: "
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 94 10094 N 0.2084461
## 1000 11000 N 0.2235163
## 591 10591 N 0.5050450
## 274 10274 N 0.5120659
## 390 10390 N 0.5392203
## 1204 11204 N 0.5392203
## 325 10325 N 0.5413975
## 1436 11437 N 0.5643877
## 1306 11307 N 0.6052334
## 1601 11602 N 0.6446219
## 184 10184 N 0.6850684
## 1795 11796 N 0.6941762
## 1747 11748 N 0.7083692
## 199 10199 N 0.8121699
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 94 N TRUE
## 1000 N TRUE
## 591 Y FALSE
## 274 Y FALSE
## 390 Y FALSE
## 1204 Y FALSE
## 325 Y FALSE
## 1436 Y FALSE
## 1306 Y FALSE
## 1601 Y FALSE
## 184 Y FALSE
## 1795 Y FALSE
## 1747 Y FALSE
## 199 Y FALSE
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 94 0.2084461
## 1000 0.2235163
## 591 0.5050450
## 274 0.5120659
## 390 0.5392203
## 1204 0.5392203
## 325 0.5413975
## 1436 0.5643877
## 1306 0.6052334
## 1601 0.6446219
## 184 0.6850684
## 1795 0.6941762
## 1747 0.7083692
## 199 0.8121699
## sold.fctr.predict.RFE.X.glmnet.accurate
## 94 TRUE
## 1000 TRUE
## 591 FALSE
## 274 FALSE
## 390 FALSE
## 1204 FALSE
## 325 FALSE
## 1436 FALSE
## 1306 FALSE
## 1601 FALSE
## 184 FALSE
## 1795 FALSE
## 1747 FALSE
## 199 FALSE
## sold.fctr.predict.RFE.X.glmnet.error .label
## 94 0.000000000 10094
## 1000 0.000000000 11000
## 591 0.005044997 10591
## 274 0.012065941 10274
## 390 0.039220348 10390
## 1204 0.039220348 11204
## 325 0.041397491 10325
## 1436 0.064387737 11437
## 1306 0.105233392 11307
## 1601 0.144621911 11602
## 184 0.185068396 10184
## 1795 0.194176209 11796
## 1747 0.208369237 11748
## 199 0.312169868 10199
## [1] "Inaccurate: "
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 1704 11705 Y 0.02766410
## 1523 11524 Y 0.03407404
## 935 10935 Y 0.06072571
## 1217 11217 Y 0.06427596
## 1816 11817 Y 0.06769666
## 332 10332 Y 0.07184129
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 1704 N FALSE
## 1523 N FALSE
## 935 N FALSE
## 1217 N FALSE
## 1816 N FALSE
## 332 N FALSE
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 1704 0.9723359
## 1523 0.9659260
## 935 0.9392743
## 1217 0.9357240
## 1816 0.9323033
## 332 0.9281587
## sold.fctr.predict.RFE.X.glmnet.accurate
## 1704 FALSE
## 1523 FALSE
## 935 FALSE
## 1217 FALSE
## 1816 FALSE
## 332 FALSE
## sold.fctr.predict.RFE.X.glmnet.error
## 1704 -0.4723359
## 1523 -0.4659260
## 935 -0.4392743
## 1217 -0.4357240
## 1816 -0.4323033
## 332 -0.4281587
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 161 10161 Y 0.09057689
## 1569 11570 Y 0.15849882
## 101 10101 Y 0.21870126
## 1190 11190 Y 0.40746370
## 325 10325 N 0.54139749
## 1052 11052 N 0.69622842
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 161 N FALSE
## 1569 N FALSE
## 101 N FALSE
## 1190 N FALSE
## 325 Y FALSE
## 1052 Y FALSE
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 161 0.9094231
## 1569 0.8415012
## 101 0.7812987
## 1190 0.5925363
## 325 0.5413975
## 1052 0.6962284
## sold.fctr.predict.RFE.X.glmnet.accurate
## 161 FALSE
## 1569 FALSE
## 101 FALSE
## 1190 FALSE
## 325 FALSE
## 1052 FALSE
## sold.fctr.predict.RFE.X.glmnet.error
## 161 -0.40942311
## 1569 -0.34150118
## 101 -0.28129874
## 1190 -0.09253630
## 325 0.04139749
## 1052 0.19622842
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 628 10628 N 0.8408622
## 504 10504 N 0.8453152
## 1690 11691 N 0.8489823
## 491 10491 N 0.8497763
## 1643 11644 N 0.8734995
## 1470 11471 N 0.9334156
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 628 Y FALSE
## 504 Y FALSE
## 1690 Y FALSE
## 491 Y FALSE
## 1643 Y FALSE
## 1470 Y FALSE
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 628 0.8408622
## 504 0.8453152
## 1690 0.8489823
## 491 0.8497763
## 1643 0.8734995
## 1470 0.9334156
## sold.fctr.predict.RFE.X.glmnet.accurate
## 628 FALSE
## 504 FALSE
## 1690 FALSE
## 491 FALSE
## 1643 FALSE
## 1470 FALSE
## sold.fctr.predict.RFE.X.glmnet.error
## 628 0.3408622
## 504 0.3453152
## 1690 0.3489823
## 491 0.3497763
## 1643 0.3734995
## 1470 0.4334156
glb_ctgry_df <- merge(glb_ctgry_df,
myget_category_stats(obs_df = glb_fitobs_df, mdl_id = glb_sel_mdl_id, label = "fit"),
by = glb_category_var, all = TRUE)## Loading required package: lazyevalrow.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
glb_ctgry_df <- merge(glb_ctgry_df,
myget_category_stats(obs_df=glb_OOBobs_df, mdl_id=glb_sel_mdl_id, label="OOB"),
by=glb_category_var, all=TRUE)
row.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
if (any(grepl("OOB", glb_model_evl_criteria)))
print(orderBy(~-err.abs.OOB.mean, glb_ctgry_df)) else
print(orderBy(~-err.abs.fit.mean, glb_ctgry_df))## prdl.descr.my.fctr .n.Tst .n.OOB.x .freqRatio.Tst
## iPadminiRetina#1 iPadminiRetina#1 NA 2 NA
## iPad5#0 iPad5#0 NA 1 NA
## iPad2#0 iPad2#0 83 93 0.10401003
## iPad1#1 iPad1#1 42 47 0.05263158
## Unknown#1 Unknown#1 47 52 0.05889724
## iPad2#1 iPad2#1 71 79 0.08897243
## iPad1#0 iPad1#0 46 52 0.05764411
## iPadmini#1 iPadmini#1 46 52 0.05764411
## iPad3#0 iPad3#0 30 34 0.03759398
## iPadmini#0 iPadmini#0 65 73 0.08145363
## Unknown#0 Unknown#0 45 50 0.05639098
## iPad4#0 iPad4#0 29 33 0.03634085
## iPadmini3#0 iPadmini3#0 29 33 0.03634085
## iPadmini2#1 iPadmini2#1 21 24 0.02631579
## iPad3#1 iPad3#1 25 28 0.03132832
## iPadAir2#0 iPadAir2#0 47 52 0.05889724
## iPadAir2#1 iPadAir2#1 15 17 0.01879699
## iPadAir#0 iPadAir#0 41 46 0.05137845
## iPad4#1 iPad4#1 39 44 0.04887218
## iPadmini2#0 iPadmini2#0 35 39 0.04385965
## iPadmini3#1 iPadmini3#1 9 10 0.01127820
## iPadminiRetina#0 iPadminiRetina#0 NA 2 NA
## iPadAir#1 iPadAir#1 33 37 0.04135338
## .freqRatio.OOB err.abs.fit.sum err.abs.fit.mean .n.fit
## iPadminiRetina#1 0.002222222 NA NA NA
## iPad5#0 0.001111111 NA NA NA
## iPad2#0 0.103333333 9.3686815 0.2129246 44
## iPad1#1 0.052222222 19.9350980 0.3559839 56
## Unknown#1 0.057777778 12.1033794 0.3667691 33
## iPad2#1 0.087777778 22.2690424 0.3181292 70
## iPad1#0 0.057777778 23.4721073 0.3353158 70
## iPadmini#1 0.057777778 17.2037432 0.3308412 52
## iPad3#0 0.037777778 9.4308076 0.2548867 37
## iPadmini#0 0.081111111 34.6998466 0.3469985 100
## Unknown#0 0.055555556 26.1411829 0.3788577 69
## iPad4#0 0.036666667 13.1137754 0.3278444 40
## iPadmini3#0 0.036666667 8.9633487 0.1991855 45
## iPadmini2#1 0.026666667 4.6215934 0.3555072 13
## iPad3#1 0.031111111 18.4753679 0.3421364 54
## iPadAir2#0 0.057777778 20.9263375 0.2790178 75
## iPadAir2#1 0.018888889 6.6020765 0.2445214 27
## iPadAir#0 0.051111111 13.3454357 0.2839454 47
## iPad4#1 0.048888889 7.8246254 0.1956156 40
## iPadmini2#0 0.043333333 12.1895562 0.3932115 31
## iPadmini3#1 0.011111111 0.2077954 0.1038977 2
## iPadminiRetina#0 0.002222222 2.1227793 0.5306948 4
## iPadAir#1 0.041111111 11.9615695 0.2392314 50
## err.abs.OOB.sum err.abs.OOB.mean .n.OOB.y
## iPadminiRetina#1 0.9496705 0.4748353 2
## iPad5#0 0.4579943 0.4579943 1
## iPad2#0 35.8776572 0.3857813 93
## iPad1#1 17.2801034 0.3676618 47
## Unknown#1 19.0108650 0.3655936 52
## iPad2#1 28.7900280 0.3644307 79
## iPad1#0 18.5486352 0.3567045 52
## iPadmini#1 18.2005569 0.3500107 52
## iPad3#0 11.6504459 0.3426602 34
## iPadmini#0 24.6734074 0.3379919 73
## Unknown#0 16.7861405 0.3357228 50
## iPad4#0 11.0562541 0.3350380 33
## iPadmini3#0 10.9791835 0.3327025 33
## iPadmini2#1 7.9783124 0.3324297 24
## iPad3#1 9.1650063 0.3273217 28
## iPadAir2#0 16.6505538 0.3202030 52
## iPadAir2#1 5.2910291 0.3112370 17
## iPadAir#0 14.3134845 0.3111627 46
## iPad4#1 13.4125718 0.3048312 44
## iPadmini2#0 11.7336648 0.3008632 39
## iPadmini3#1 2.7265614 0.2726561 10
## iPadminiRetina#0 0.5069519 0.2534760 2
## iPadAir#1 9.2942141 0.2511950 37print(colSums(glb_ctgry_df[, -grep(glb_category_var, names(glb_ctgry_df))]))## .n.Tst .n.OOB.x .freqRatio.Tst .freqRatio.OOB
## NA 900.000000 NA 1.000000
## err.abs.fit.sum err.abs.fit.mean .n.fit err.abs.OOB.sum
## NA NA NA 305.333292
## err.abs.OOB.mean .n.OOB.y
## 7.792503 900.000000write.csv(glb_OOBobs_df[, c(glb_id_var,
grep(glb_rsp_var, names(glb_OOBobs_df), fixed=TRUE, value=TRUE))],
paste0(gsub(".", "_", paste0(glb_out_pfx, glb_sel_mdl_id), fixed=TRUE),
"_OOBobs.csv"), row.names=FALSE)
glb_chunks_df <- myadd_chunk(glb_chunks_df, "fit.models", major.inc=FALSE)## label step_major step_minor label_minor bgn end elapsed
## 12 fit.models 7 2 2 180.214 205.309 25.095
## 13 fit.models 7 3 3 205.309 NA NA# if (sum(is.na(glb_allobs_df$D.P.http)) > 0)
# stop("fit.models_3: Why is this happening ?")
#stop(here"); glb_to_sav()
sync_glb_obs_df <- function() {
# Merge or cbind ?
for (col in setdiff(names(glb_fitobs_df), names(glb_trnobs_df)))
glb_trnobs_df[glb_trnobs_df$.lcn == "Fit", col] <<- glb_fitobs_df[, col]
for (col in setdiff(names(glb_fitobs_df), names(glb_allobs_df)))
glb_allobs_df[glb_allobs_df$.lcn == "Fit", col] <<- glb_fitobs_df[, col]
if (all(is.na(glb_newobs_df[, glb_rsp_var])))
for (col in setdiff(names(glb_OOBobs_df), names(glb_trnobs_df)))
glb_trnobs_df[glb_trnobs_df$.lcn == "OOB", col] <<- glb_OOBobs_df[, col]
for (col in setdiff(names(glb_OOBobs_df), names(glb_allobs_df)))
glb_allobs_df[glb_allobs_df$.lcn == "OOB", col] <<- glb_OOBobs_df[, col]
}
sync_glb_obs_df()
print(setdiff(names(glb_newobs_df), names(glb_allobs_df)))## character(0)if (glb_save_envir)
save(glb_feats_df,
glb_allobs_df, #glb_trnobs_df, glb_fitobs_df, glb_OOBobs_df, glb_newobs_df,
glb_models_df, dsp_models_df, glb_models_lst, glb_sel_mdl, glb_sel_mdl_id,
glb_model_type,
file=paste0(glb_out_pfx, "selmdl_dsk.RData"))
#load(paste0(glb_out_pfx, "selmdl_dsk.RData"))
rm(ret_lst)## Warning in rm(ret_lst): object 'ret_lst' not foundreplay.petrisim(pn=glb_analytics_pn,
replay.trans=(glb_analytics_avl_objs <- c(glb_analytics_avl_objs,
"model.selected")), flip_coord=TRUE)## time trans "bgn " "fit.data.training.all " "predict.data.new " "end "
## 0.0000 multiple enabled transitions: data.training.all data.new model.selected firing: data.training.all
## 1.0000 1 2 1 0 0
## 1.0000 multiple enabled transitions: data.training.all data.new model.selected model.final data.training.all.prediction firing: data.new
## 2.0000 2 1 1 1 0
## 2.0000 multiple enabled transitions: data.training.all data.new model.selected model.final data.training.all.prediction data.new.prediction firing: model.selected
## 3.0000 3 0 2 1 0
glb_chunks_df <- myadd_chunk(glb_chunks_df, "fit.data.training", major.inc=TRUE)## label step_major step_minor label_minor bgn end
## 13 fit.models 7 3 3 205.309 210.915
## 14 fit.data.training 8 0 0 210.916 NA
## elapsed
## 13 5.606
## 14 NAStep 8.0: fit data training
#load(paste0(glb_inp_pfx, "dsk.RData"))
if (!is.null(glb_fin_mdl_id) && (glb_fin_mdl_id %in% names(glb_models_lst))) {
warning("Final model same as user selected model")
glb_fin_mdl <- glb_models_lst[[glb_fin_mdl_id]]
} else if (nrow(glb_fitobs_df) + length(glb_obsfit_outliers) == nrow(glb_trnobs_df)) {
warning("Final model same as glb_sel_mdl_id")
glb_fin_mdl_id <- paste0("Final.", glb_sel_mdl_id)
glb_fin_mdl <- glb_sel_mdl
glb_models_lst[[glb_fin_mdl_id]] <- glb_fin_mdl
} else {
if (grepl("RFE", glb_sel_mdl_id) || grepl("RFE", glb_mdl_ensemble)) {
print("***************")
print("RFE indep_vars should be based on glb_trnobs_df. Also, outliers in OOB.\nNot implemented yet")
print("***************")
}
if (grepl("Ensemble", glb_sel_mdl_id)) {
# Find which models are relevant
mdlimp_df <- subset(myget_feats_importance(glb_sel_mdl), importance > 5)
# Fit selected models on glb_trnobs_df
for (mdl_id in gsub(".prob", "",
gsub(glb_rsp_var_out, "", row.names(mdlimp_df), fixed=TRUE),
fixed=TRUE)) {
mdl_id_components <- unlist(strsplit(mdl_id, "[.]"))
ret_lst <-
myfit_mdl(mdl_id=paste0(c(head(mdl_id_components, -1), "Train"),
collapse="."),
model_method=tail(mdl_id_components, 1),
indep_vars_vctr=trim(unlist(strsplit(
glb_models_df[glb_models_df$id == mdl_id, "feats"], "[,]"))),
model_type=glb_model_type,
rsp_var=glb_rsp_var, rsp_var_out=glb_rsp_var_out,
fit_df=glb_trnobs_df, OOB_df=NULL,
n_cv_folds=glb_rcv_n_folds, tune_models_df=glb_tune_models_df)
glb_trnobs_df <- glb_get_predictions(df=glb_trnobs_df,
mdl_id=tail(glb_models_df$id, 1),
rsp_var_out=glb_rsp_var_out,
prob_threshold_def=subset(glb_models_df,
mdl_id == mdl_id)$opt.prob.threshold.OOB)
glb_newobs_df <- glb_get_predictions(df=glb_newobs_df,
mdl_id=tail(glb_models_df$id, 1),
rsp_var_out=glb_rsp_var_out,
prob_threshold_def=subset(glb_models_df,
mdl_id == mdl_id)$opt.prob.threshold.OOB)
}
}
# "Final" model
if ((model_method <- glb_sel_mdl$method) == "custom")
# get actual method from the mdl_id
model_method <- tail(unlist(strsplit(glb_sel_mdl_id, "[.]")), 1)
if (grepl("Ensemble", glb_sel_mdl_id)) {
# Find which models are relevant
mdlimp_df <- subset(myget_feats_importance(glb_sel_mdl), importance > 5)
if (glb_is_classification && glb_is_binomial)
indep_vars_vctr <- gsub("(.*)\\.(.*)\\.prob", "\\1\\.Train\\.\\2\\.prob",
row.names(mdlimp_df)) else
indep_vars_vctr <- gsub("(.*)\\.(.*)", "\\1\\.Train\\.\\2",
row.names(mdlimp_df))
} else indep_vars_vctr <-
trim(unlist(strsplit(glb_models_df[glb_models_df$id ==
glb_sel_mdl_id
, "feats"], "[,]")))
# Discontinuing use of tune_finmdl_df;
# since final model needs to be cved on glb_trnobs_df
tune_finmdl_df <- NULL
if (nrow(glb_sel_mdl$bestTune) > 0) {
for (param in names(glb_sel_mdl$bestTune)) {
#print(sprintf("param: %s", param))
if (glb_sel_mdl$bestTune[1, param] != "none")
tune_finmdl_df <- rbind(tune_finmdl_df,
data.frame(parameter=param,
min=glb_sel_mdl$bestTune[1, param],
max=glb_sel_mdl$bestTune[1, param],
by=1)) # by val does not matter
}
}
# Sync with parameters in mydsutils.R
#stop(here"); glb_to_sav(); glb_models_lst <- sav_models_lst; glb_models_df <- sav_models_df
if (!is.null(glb_preproc_methods) &&
((match_pos <- regexpr(gsub(".", "\\.", paste(glb_preproc_methods, collapse="|"),
fixed=TRUE), glb_sel_mdl_id)) != -1))
ths_preProcess <- str_sub(glb_sel_mdl_id, match_pos,
match_pos + attr(match_pos, "match.length") - 1) else
ths_preProcess <- NULL
fit_trnobs_df <- if (is.null(glb_obstrn_outliers)) glb_trnobs_df else
glb_trnobs_df[!(glb_trnobs_df[, glb_id_var] %in% glb_obstrn_outliers), ]
# Force fitting of Final.glm to identify outliers
method_vctr <- unique(c("glm", tail(unlist(strsplit(glb_sel_mdl_id, "[.]")), 1)))
for (method in method_vctr) {
#source("caret_nominalTrainWorkflow.R")
ret_lst <- myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst = list(
id.prefix = ifelse(grepl("Ensemble", glb_sel_mdl_id), "Final.Ensemble",
"Final"),
type = glb_model_type, trainControl.method = "repeatedcv",
trainControl.number = glb_rcv_n_folds,
trainControl.repeats = glb_rcv_n_repeats,
train.method = method,
train.preProcess = ths_preProcess)),
indep_vars = indep_vars_vctr, rsp_var = glb_rsp_var,
fit_df = fit_trnobs_df, OOB_df = NULL)
if ((length(method_vctr) == 1) || (method != "glm")) {
glb_fin_mdl <- glb_models_lst[[length(glb_models_lst)]]
glb_fin_mdl_id <- glb_models_df[length(glb_models_lst), "id"]
}
}
}## [1] "***************"
## [1] "RFE indep_vars should be based on glb_trnobs_df. Also, outliers in OOB.\nNot implemented yet"
## [1] "***************"
## [1] "fitting model: Final.glm"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr"
## + Fold1.Rep1: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold1.Rep1: parameter=none
## + Fold2.Rep1: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold2.Rep1: parameter=none
## + Fold3.Rep1: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold3.Rep1: parameter=none
## + Fold1.Rep2: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold1.Rep2: parameter=none
## + Fold2.Rep2: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold2.Rep2: parameter=none
## + Fold3.Rep2: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold3.Rep2: parameter=none
## + Fold1.Rep3: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold1.Rep3: parameter=none
## + Fold2.Rep3: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold2.Rep3: parameter=none
## + Fold3.Rep3: parameter=none## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading## - Fold3.Rep3: parameter=none
## Aggregating results
## Fitting final model on full training set## Warning: not plotting observations with leverage one:
## 1057
## Warning: not plotting observations with leverage one:
## 1057
##
## Call:
## NULL
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.7626 -0.7251 -0.2595 0.6352 3.6592
##
## Coefficients: (13 not defined because of singularities)
## Estimate Std. Error z value
## (Intercept) 1.920e+00 5.816e-01 3.301
## .rnorm 6.403e-02 6.060e-02 1.057
## biddable 1.501e+00 1.379e-01 10.881
## cellular.fctr1 6.236e-02 2.067e-01 0.302
## cellular.fctrUnknown -5.927e-01 3.242e-01 -1.828
## color.fctrGold -5.585e-01 4.585e-01 -1.218
## `color.fctrSpace Gray` -1.521e-01 2.540e-01 -0.599
## color.fctrUnknown -1.369e-01 1.697e-01 -0.807
## color.fctrWhite -1.934e-01 1.849e-01 -1.046
## `condition.fctrFor parts or not working` -5.884e-01 2.157e-01 -2.728
## `condition.fctrManufacturer refurbished` 1.992e-01 4.475e-01 0.445
## condition.fctrNew 2.299e-01 2.447e-01 0.940
## `condition.fctrNew other (see details)` 3.979e-01 3.266e-01 1.218
## `condition.fctrSeller refurbished` -5.919e-01 2.790e-01 -2.121
## `prdl.descr.my.fctrUnknown#1` 3.858e-01 3.858e-01 1.000
## `prdl.descr.my.fctriPad1#0` -2.969e-01 3.868e-01 -0.768
## `prdl.descr.my.fctriPad1#1` -1.657e-01 3.930e-01 -0.422
## `prdl.descr.my.fctriPad2#0` 2.981e-01 3.792e-01 0.786
## `prdl.descr.my.fctriPad2#1` 8.407e-02 3.756e-01 0.224
## `prdl.descr.my.fctriPad3#0` 7.254e-01 4.485e-01 1.618
## `prdl.descr.my.fctriPad3#1` 6.859e-01 4.249e-01 1.614
## `prdl.descr.my.fctriPad4#0` 1.358e+00 4.475e-01 3.034
## `prdl.descr.my.fctriPad4#1` 3.707e-01 4.618e-01 0.803
## `prdl.descr.my.fctriPad5#0` 2.738e+00 6.523e+02 0.004
## `prdl.descr.my.fctriPadAir#0` 2.115e+00 4.514e-01 4.685
## `prdl.descr.my.fctriPadAir#1` 1.212e+00 4.706e-01 2.576
## `prdl.descr.my.fctriPadAir2#0` 3.321e+00 5.081e-01 6.536
## `prdl.descr.my.fctriPadAir2#1` 2.926e+00 6.106e-01 4.793
## `prdl.descr.my.fctriPadmini#0` 3.836e-01 3.668e-01 1.046
## `prdl.descr.my.fctriPadmini#1` 3.419e-01 4.008e-01 0.853
## `prdl.descr.my.fctriPadmini2#0` 1.550e+00 4.601e-01 3.370
## `prdl.descr.my.fctriPadmini2#1` 1.053e+00 5.420e-01 1.942
## `prdl.descr.my.fctriPadmini3#0` 1.962e+00 5.119e-01 3.832
## `prdl.descr.my.fctriPadmini3#1` 1.966e+00 9.957e-01 1.974
## `prdl.descr.my.fctriPadminiRetina#0` 2.744e+00 9.930e-01 2.763
## `prdl.descr.my.fctriPadminiRetina#1` 2.915e+00 1.515e+00 1.924
## startprice -1.166e-02 9.261e-04 -12.594
## storage.fctr16 -1.215e+00 4.224e-01 -2.877
## storage.fctr32 -1.164e+00 4.324e-01 -2.693
## storage.fctr64 -6.341e-01 4.226e-01 -1.500
## storage.fctrUnknown -7.814e-01 5.294e-01 -1.476
## `cellular.fctr0:carrier.fctrNone` NA NA NA
## `cellular.fctr1:carrier.fctrNone` NA NA NA
## `cellular.fctrUnknown:carrier.fctrNone` NA NA NA
## `cellular.fctr0:carrier.fctrOther` NA NA NA
## `cellular.fctr1:carrier.fctrOther` 1.175e+01 3.726e+02 0.032
## `cellular.fctrUnknown:carrier.fctrOther` NA NA NA
## `cellular.fctr0:carrier.fctrSprint` NA NA NA
## `cellular.fctr1:carrier.fctrSprint` 6.699e-01 5.461e-01 1.227
## `cellular.fctrUnknown:carrier.fctrSprint` NA NA NA
## `cellular.fctr0:carrier.fctrT-Mobile` NA NA NA
## `cellular.fctr1:carrier.fctrT-Mobile` 4.616e-01 7.352e-01 0.628
## `cellular.fctrUnknown:carrier.fctrT-Mobile` NA NA NA
## `cellular.fctr0:carrier.fctrUnknown` NA NA NA
## `cellular.fctr1:carrier.fctrUnknown` 7.421e-02 3.438e-01 0.216
## `cellular.fctrUnknown:carrier.fctrUnknown` NA NA NA
## `cellular.fctr0:carrier.fctrVerizon` NA NA NA
## `cellular.fctr1:carrier.fctrVerizon` 2.627e-01 2.970e-01 0.885
## `cellular.fctrUnknown:carrier.fctrVerizon` NA NA NA
## Pr(>|z|)
## (Intercept) 0.000964 ***
## .rnorm 0.290665
## biddable < 2e-16 ***
## cellular.fctr1 0.762903
## cellular.fctrUnknown 0.067488 .
## color.fctrGold 0.223205
## `color.fctrSpace Gray` 0.549354
## color.fctrUnknown 0.419683
## color.fctrWhite 0.295712
## `condition.fctrFor parts or not working` 0.006379 **
## `condition.fctrManufacturer refurbished` 0.656269
## condition.fctrNew 0.347374
## `condition.fctrNew other (see details)` 0.223123
## `condition.fctrSeller refurbished` 0.033890 *
## `prdl.descr.my.fctrUnknown#1` 0.317385
## `prdl.descr.my.fctriPad1#0` 0.442700
## `prdl.descr.my.fctriPad1#1` 0.673209
## `prdl.descr.my.fctriPad2#0` 0.431895
## `prdl.descr.my.fctriPad2#1` 0.822913
## `prdl.descr.my.fctriPad3#0` 0.105758
## `prdl.descr.my.fctriPad3#1` 0.106496
## `prdl.descr.my.fctriPad4#0` 0.002415 **
## `prdl.descr.my.fctriPad4#1` 0.422145
## `prdl.descr.my.fctriPad5#0` 0.996651
## `prdl.descr.my.fctriPadAir#0` 2.80e-06 ***
## `prdl.descr.my.fctriPadAir#1` 0.009996 **
## `prdl.descr.my.fctriPadAir2#0` 6.31e-11 ***
## `prdl.descr.my.fctriPadAir2#1` 1.65e-06 ***
## `prdl.descr.my.fctriPadmini#0` 0.295626
## `prdl.descr.my.fctriPadmini#1` 0.393566
## `prdl.descr.my.fctriPadmini2#0` 0.000752 ***
## `prdl.descr.my.fctriPadmini2#1` 0.052093 .
## `prdl.descr.my.fctriPadmini3#0` 0.000127 ***
## `prdl.descr.my.fctriPadmini3#1` 0.048371 *
## `prdl.descr.my.fctriPadminiRetina#0` 0.005723 **
## `prdl.descr.my.fctriPadminiRetina#1` 0.054388 .
## startprice < 2e-16 ***
## storage.fctr16 0.004009 **
## storage.fctr32 0.007082 **
## storage.fctr64 0.133487
## storage.fctrUnknown 0.139936
## `cellular.fctr0:carrier.fctrNone` NA
## `cellular.fctr1:carrier.fctrNone` NA
## `cellular.fctrUnknown:carrier.fctrNone` NA
## `cellular.fctr0:carrier.fctrOther` NA
## `cellular.fctr1:carrier.fctrOther` 0.974840
## `cellular.fctrUnknown:carrier.fctrOther` NA
## `cellular.fctr0:carrier.fctrSprint` NA
## `cellular.fctr1:carrier.fctrSprint` 0.219910
## `cellular.fctrUnknown:carrier.fctrSprint` NA
## `cellular.fctr0:carrier.fctrT-Mobile` NA
## `cellular.fctr1:carrier.fctrT-Mobile` 0.530121
## `cellular.fctrUnknown:carrier.fctrT-Mobile` NA
## `cellular.fctr0:carrier.fctrUnknown` NA
## `cellular.fctr1:carrier.fctrUnknown` 0.829079
## `cellular.fctrUnknown:carrier.fctrUnknown` NA
## `cellular.fctr0:carrier.fctrVerizon` NA
## `cellular.fctr1:carrier.fctrVerizon` 0.376416
## `cellular.fctrUnknown:carrier.fctrVerizon` NA
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 2566.7 on 1858 degrees of freedom
## Residual deviance: 1675.7 on 1813 degrees of freedom
## AIC: 1767.7
##
## Number of Fisher Scoring iterations: 12
##
## [1] " calling mypredict_mdl for fit:"## Warning in predict.lm(object, newdata, se.fit, scale = 1, type =
## ifelse(type == : prediction from a rank-deficient fit may be misleading
## threshold f.score
## 1 0.0 0.6325855
## 2 0.1 0.6843393
## 3 0.2 0.7228585
## 4 0.3 0.7528557
## 5 0.4 0.7692308
## 6 0.5 0.7748184
## 7 0.6 0.7813485
## 8 0.7 0.7678812
## 9 0.8 0.6535254
## 10 0.9 0.3775701
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.6000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Final.glm.N sold.fctr.predict.Final.glm.Y
## 1 N 892 107
## 2 Y 240 620
## Prediction
## Reference N Y
## N 892 107
## Y 240 620
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.133405e-01 6.204991e-01 7.948702e-01 8.308164e-01 5.373857e-01
## AccuracyPValue McnemarPValue
## 1.403435e-137 1.379130e-12## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## id
## 1 Final.glm
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 1 2.573 0.141
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8665514 0.6 0.7813485 0.7939766
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7948702 0.8308164 0.5835849 0.8133405
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01020258 0.02104005
## [1] "fitting model: Final.glmnet"
## [1] " indep_vars: startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr"
## Aggregating results
## Selecting tuning parameters
## Fitting alpha = 0.55, lambda = 0.000254 on full training set## Warning in myfit_mdl(mdl_specs_lst = myinit_mdl_specs_lst(mdl_specs_lst =
## list(id.prefix = ifelse(grepl("Ensemble", : model's bestTune found at an
## extreme of tuneGrid for parameter: lambda
## Length Class Mode
## a0 86 -none- numeric
## beta 4988 dgCMatrix S4
## df 86 -none- numeric
## dim 2 -none- numeric
## lambda 86 -none- numeric
## dev.ratio 86 -none- numeric
## nulldev 1 -none- numeric
## npasses 1 -none- numeric
## jerr 1 -none- numeric
## offset 1 -none- logical
## classnames 2 -none- character
## call 5 -none- call
## nobs 1 -none- numeric
## lambdaOpt 1 -none- numeric
## xNames 58 -none- character
## problemType 1 -none- character
## tuneValue 2 data.frame list
## obsLevels 2 -none- character
## [1] "min lambda > lambdaOpt:"
## (Intercept)
## 1.818972664
## .rnorm
## 0.062273775
## biddable
## 1.512131444
## cellular.fctr1
## 0.061578533
## cellular.fctrUnknown
## -0.298481843
## color.fctrGold
## -0.522188744
## color.fctrSpace Gray
## -0.139152442
## color.fctrUnknown
## -0.124440067
## color.fctrWhite
## -0.180545444
## condition.fctrFor parts or not working
## -0.564130273
## condition.fctrManufacturer refurbished
## 0.184606297
## condition.fctrNew
## 0.209244111
## condition.fctrNew other (see details)
## 0.386905808
## condition.fctrSeller refurbished
## -0.583499097
## prdl.descr.my.fctrUnknown#1
## 0.305449477
## prdl.descr.my.fctriPad1#0
## -0.366624726
## prdl.descr.my.fctriPad1#1
## -0.239913909
## prdl.descr.my.fctriPad2#0
## 0.202850449
## prdl.descr.my.fctriPad2#1
## -0.005648497
## prdl.descr.my.fctriPad3#0
## 0.614988893
## prdl.descr.my.fctriPad3#1
## 0.574362350
## prdl.descr.my.fctriPad4#0
## 1.238185814
## prdl.descr.my.fctriPad4#1
## 0.249048428
## prdl.descr.my.fctriPad5#0
## 1.754420812
## prdl.descr.my.fctriPadAir#0
## 1.980458189
## prdl.descr.my.fctriPadAir#1
## 1.075235285
## prdl.descr.my.fctriPadAir2#0
## 3.139130080
## prdl.descr.my.fctriPadAir2#1
## 2.745604232
## prdl.descr.my.fctriPadmini#0
## 0.277780402
## prdl.descr.my.fctriPadmini#1
## 0.236907249
## prdl.descr.my.fctriPadmini2#0
## 1.420399913
## prdl.descr.my.fctriPadmini2#1
## 0.922163407
## prdl.descr.my.fctriPadmini3#0
## 1.806630231
## prdl.descr.my.fctriPadmini3#1
## 1.780439114
## prdl.descr.my.fctriPadminiRetina#0
## 2.594507360
## prdl.descr.my.fctriPadminiRetina#1
## 2.741191910
## startprice
## -0.011346194
## storage.fctr16
## -1.073518533
## storage.fctr32
## -1.027931875
## storage.fctr64
## -0.506969948
## storage.fctrUnknown
## -0.645247952
## cellular.fctr1:carrier.fctrOther
## 4.056406785
## cellular.fctr1:carrier.fctrSprint
## 0.665432931
## cellular.fctr1:carrier.fctrT-Mobile
## 0.447268039
## cellular.fctr1:carrier.fctrUnknown
## 0.068115574
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.328074731
## cellular.fctr1:carrier.fctrVerizon
## 0.254345420
## [1] "max lambda < lambdaOpt:"
## (Intercept)
## 1.82728327
## .rnorm
## 0.06242624
## biddable
## 1.51120758
## cellular.fctr1
## 0.06160843
## cellular.fctrUnknown
## -0.29686685
## color.fctrGold
## -0.52528302
## color.fctrSpace Gray
## -0.14037101
## color.fctrUnknown
## -0.12549578
## color.fctrWhite
## -0.18164400
## condition.fctrFor parts or not working
## -0.56608177
## condition.fctrManufacturer refurbished
## 0.18576830
## condition.fctrNew
## 0.21094054
## condition.fctrNew other (see details)
## 0.38779076
## condition.fctrSeller refurbished
## -0.58422113
## prdl.descr.my.fctrUnknown#1
## 0.31178050
## prdl.descr.my.fctriPad1#0
## -0.36141556
## prdl.descr.my.fctriPad1#1
## -0.23433002
## prdl.descr.my.fctriPad2#0
## 0.21013667
## prdl.descr.my.fctriPad3#0
## 0.62344066
## prdl.descr.my.fctriPad3#1
## 0.58291929
## prdl.descr.my.fctriPad4#0
## 1.24742303
## prdl.descr.my.fctriPad4#1
## 0.25840994
## prdl.descr.my.fctriPad5#0
## 1.77111430
## prdl.descr.my.fctriPadAir#0
## 1.99098152
## prdl.descr.my.fctriPadAir#1
## 1.08592171
## prdl.descr.my.fctriPadAir2#0
## 3.15343725
## prdl.descr.my.fctriPadAir2#1
## 2.75985871
## prdl.descr.my.fctriPadmini#0
## 0.28585121
## prdl.descr.my.fctriPadmini#1
## 0.24491332
## prdl.descr.my.fctriPadmini2#0
## 1.43047407
## prdl.descr.my.fctriPadmini2#1
## 0.93226725
## prdl.descr.my.fctriPadmini3#0
## 1.81872680
## prdl.descr.my.fctriPadmini3#1
## 1.79511424
## prdl.descr.my.fctriPadminiRetina#0
## 2.60617964
## prdl.descr.my.fctriPadminiRetina#1
## 2.75491972
## startprice
## -0.01137189
## storage.fctr16
## -1.08452378
## storage.fctr32
## -1.03850314
## storage.fctr64
## -0.51679349
## storage.fctrUnknown
## -0.65602396
## cellular.fctr1:carrier.fctrOther
## 4.14404816
## cellular.fctr1:carrier.fctrSprint
## 0.66590285
## cellular.fctr1:carrier.fctrT-Mobile
## 0.44850810
## cellular.fctr1:carrier.fctrUnknown
## 0.06869847
## cellular.fctrUnknown:carrier.fctrUnknown
## -0.32715773
## cellular.fctr1:carrier.fctrVerizon
## 0.25527076
## [1] " calling mypredict_mdl for fit:"
## threshold f.score
## 1 0.0 0.6325855
## 2 0.1 0.6845966
## 3 0.2 0.7231121
## 4 0.3 0.7524650
## 5 0.4 0.7680827
## 6 0.5 0.7762280
## 7 0.6 0.7831021
## 8 0.7 0.7715054
## 9 0.8 0.6494297
## 10 0.9 0.3656927
## 11 1.0 0.0000000## [1] "Classifier Probability Threshold: 0.6000 to maximize f.score.fit"
## sold.fctr sold.fctr.predict.Final.glmnet.N
## 1 N 894
## 2 Y 239
## sold.fctr.predict.Final.glmnet.Y
## 1 105
## 2 621
## Prediction
## Reference N Y
## N 894 105
## Y 239 621
## Accuracy Kappa AccuracyLower AccuracyUpper AccuracyNull
## 8.149543e-01 6.237493e-01 7.965405e-01 8.323684e-01 5.373857e-01
## AccuracyPValue McnemarPValue
## 2.615624e-139 7.451752e-13## Warning in mypredict_mdl(mdl, df = obs_df, rsp_var, rsp_var_out,
## mdl_specs_lst[["id"]], : Expecting 1 metric: ; recd: Accuracy, Kappa;
## retaining Accuracy only
## [1] " calling mypredict_mdl for OOB:"
## id
## 1 Final.glmnet
## feats
## 1 startprice,biddable,prdl.descr.my.fctr,condition.fctr,cellular.fctr,storage.fctr,.rnorm,color.fctr,cellular.fctr:carrier.fctr
## max.nTuningRuns min.elapsedtime.everything min.elapsedtime.final
## 1 25 3.569 0.197
## max.auc.fit opt.prob.threshold.fit max.f.score.fit max.Accuracy.fit
## 1 0.8665293 0.6 0.7831021 0.7930797
## max.AccuracyLower.fit max.AccuracyUpper.fit max.Kappa.fit .fit
## 1 0.7965405 0.8323684 0.5818559 0.8149543
## max.AccuracySD.fit max.KappaSD.fit
## 1 0.01070114 0.02193226rm(ret_lst)
glb_chunks_df <- myadd_chunk(glb_chunks_df, "fit.data.training", major.inc=FALSE)## label step_major step_minor label_minor bgn end
## 14 fit.data.training 8 0 0 210.916 223.101
## 15 fit.data.training 8 1 1 223.102 NA
## elapsed
## 14 12.185
## 15 NA#stop(here"); glb_to_sav()
if (glb_is_classification && glb_is_binomial)
prob_threshold <- glb_models_df[glb_models_df$id == glb_sel_mdl_id,
"opt.prob.threshold.OOB"] else
prob_threshold <- NULL
if (grepl("Ensemble", glb_fin_mdl_id)) {
# Get predictions for each model in ensemble; Outliers that have been moved to OOB might not have been predicted yet
for (mdl_id in glb_mdl_ensemble) {
glb_trnobs_df <- glb_get_predictions(df=glb_trnobs_df, mdl_id=mdl_id,
rsp_var_out=glb_rsp_var_out,
prob_threshold_def=prob_threshold)
glb_newobs_df <- glb_get_predictions(df=glb_newobs_df, mdl_id=mdl_id,
rsp_var_out=glb_rsp_var_out,
prob_threshold_def=prob_threshold)
}
}
glb_trnobs_df <- glb_get_predictions(df=glb_trnobs_df, mdl_id=glb_fin_mdl_id,
rsp_var_out=glb_rsp_var_out,
prob_threshold_def=prob_threshold)## Warning in glb_get_predictions(df = glb_trnobs_df, mdl_id =
## glb_fin_mdl_id, : Using default probability threshold: 0.5glb_featsimp_df <- myget_feats_importance(mdl=glb_fin_mdl,
featsimp_df=glb_featsimp_df)
glb_featsimp_df[, paste0(glb_fin_mdl_id, ".importance")] <- glb_featsimp_df$importance
print(glb_featsimp_df)## RFE.X.glmnet.importance
## cellular.fctr1:carrier.fctrOther 38.77602
## prdl.descr.my.fctriPadAir2#0 62.34763
## prdl.descr.my.fctriPadAir2#1 55.66092
## prdl.descr.my.fctriPadminiRetina#1 38.77602
## prdl.descr.my.fctriPadminiRetina#0 69.85328
## prdl.descr.my.fctriPadAir#0 55.88955
## prdl.descr.my.fctriPadmini3#0 36.37694
## prdl.descr.my.fctriPadmini3#1 38.77602
## prdl.descr.my.fctriPad5#0 38.77602
## biddable 100.00000
## prdl.descr.my.fctriPadmini2#0 44.08210
## prdl.descr.my.fctriPad4#0 50.37100
## prdl.descr.my.fctriPadAir#1 38.77602
## prdl.descr.my.fctriPadmini2#1 38.77602
## cellular.fctr1:carrier.fctrSprint 38.77602
## prdl.descr.my.fctriPad3#0 47.81549
## prdl.descr.my.fctriPad3#1 38.77602
## cellular.fctr1:carrier.fctrT-Mobile 50.56798
## condition.fctrNew other (see details) 38.77602
## prdl.descr.my.fctrUnknown#1 38.11731
## prdl.descr.my.fctriPadmini#0 36.09147
## cellular.fctr1:carrier.fctrVerizon 41.54999
## prdl.descr.my.fctriPad4#1 0.00000
## prdl.descr.my.fctriPadmini#1 38.77602
## condition.fctrNew 38.77602
## prdl.descr.my.fctriPad2#0 63.75435
## condition.fctrManufacturer refurbished 38.77602
## cellular.fctr1:carrier.fctrUnknown 39.72591
## .rnorm 38.77602
## cellular.fctr1 43.54682
## cellular.fctr0:carrier.fctrNone 38.77602
## cellular.fctr0:carrier.fctrOther 38.77602
## cellular.fctr0:carrier.fctrSprint 38.77602
## cellular.fctr0:carrier.fctrT-Mobile 38.77602
## cellular.fctr0:carrier.fctrUnknown 38.77602
## cellular.fctr0:carrier.fctrVerizon 38.77602
## cellular.fctr1:carrier.fctrNone 38.77602
## cellular.fctrUnknown:carrier.fctrNone 38.77602
## cellular.fctrUnknown:carrier.fctrOther 38.77602
## cellular.fctrUnknown:carrier.fctrSprint 38.77602
## cellular.fctrUnknown:carrier.fctrT-Mobile 38.77602
## cellular.fctrUnknown:carrier.fctrVerizon 38.77602
## prdl.descr.my.fctriPad2#1 32.06133
## startprice 38.55010
## color.fctrUnknown 38.77602
## color.fctrSpace Gray 38.77602
## color.fctrWhite 38.77602
## prdl.descr.my.fctriPad1#1 38.77602
## cellular.fctrUnknown 28.65476
## cellular.fctrUnknown:carrier.fctrUnknown 29.05086
## prdl.descr.my.fctriPad1#0 38.77602
## storage.fctr64 41.91171
## color.fctrGold 38.77602
## condition.fctrFor parts or not working 38.77602
## condition.fctrSeller refurbished 25.06043
## storage.fctrUnknown 38.77602
## storage.fctr32 37.62001
## storage.fctr16 38.77602
## importance
## cellular.fctr1:carrier.fctrOther 100.0000000
## prdl.descr.my.fctriPadAir2#0 81.5951263
## prdl.descr.my.fctriPadAir2#1 73.9946427
## prdl.descr.my.fctriPadminiRetina#1 73.9044747
## prdl.descr.my.fctriPadminiRetina#0 71.0522837
## prdl.descr.my.fctriPadAir#0 59.1826188
## prdl.descr.my.fctriPadmini3#0 55.8403584
## prdl.descr.my.fctriPadmini3#1 55.3587999
## prdl.descr.my.fctriPad5#0 54.8753087
## biddable 50.0303408
## prdl.descr.my.fctriPadmini2#0 48.3622392
## prdl.descr.my.fctriPad4#0 44.8353463
## prdl.descr.my.fctriPadAir#1 41.7019899
## prdl.descr.my.fctriPadmini2#1 38.7403007
## cellular.fctr1:carrier.fctrSprint 33.6915454
## prdl.descr.my.fctriPad3#0 32.7924348
## prdl.descr.my.fctriPad3#1 32.0088221
## cellular.fctr1:carrier.fctrT-Mobile 29.4854703
## condition.fctrNew other (see details) 28.3163808
## prdl.descr.my.fctrUnknown#1 26.7944874
## prdl.descr.my.fctriPadmini#0 26.2764947
## cellular.fctr1:carrier.fctrVerizon 25.7566816
## prdl.descr.my.fctriPad4#1 25.7337498
## prdl.descr.my.fctriPadmini#1 25.4865208
## condition.fctrNew 24.8929146
## prdl.descr.my.fctriPad2#0 24.8220248
## condition.fctrManufacturer refurbished 24.4120678
## cellular.fctr1:carrier.fctrUnknown 22.1568866
## .rnorm 22.0400171
## cellular.fctr1 22.0254370
## cellular.fctr0:carrier.fctrNone 20.8359173
## cellular.fctr0:carrier.fctrOther 20.8359173
## cellular.fctr0:carrier.fctrSprint 20.8359173
## cellular.fctr0:carrier.fctrT-Mobile 20.8359173
## cellular.fctr0:carrier.fctrUnknown 20.8359173
## cellular.fctr0:carrier.fctrVerizon 20.8359173
## cellular.fctr1:carrier.fctrNone 20.8359173
## cellular.fctrUnknown:carrier.fctrNone 20.8359173
## cellular.fctrUnknown:carrier.fctrOther 20.8359173
## cellular.fctrUnknown:carrier.fctrSprint 20.8359173
## cellular.fctrUnknown:carrier.fctrT-Mobile 20.8359173
## cellular.fctrUnknown:carrier.fctrVerizon 20.8359173
## prdl.descr.my.fctriPad2#1 20.7799709
## startprice 20.6165515
## color.fctrUnknown 18.4227302
## color.fctrSpace Gray 18.1370645
## color.fctrWhite 17.3387891
## prdl.descr.my.fctriPad1#1 16.2551004
## cellular.fctrUnknown 15.0866653
## cellular.fctrUnknown:carrier.fctrUnknown 14.5085844
## prdl.descr.my.fctriPad1#0 13.8044660
## storage.fctr64 10.9526155
## color.fctrGold 10.7220100
## condition.fctrFor parts or not working 9.9227633
## condition.fctrSeller refurbished 9.5602685
## storage.fctrUnknown 8.2731539
## storage.fctr32 0.8844774
## storage.fctr16 0.0000000
## Final.glmnet.importance
## cellular.fctr1:carrier.fctrOther 100.0000000
## prdl.descr.my.fctriPadAir2#0 81.5951263
## prdl.descr.my.fctriPadAir2#1 73.9946427
## prdl.descr.my.fctriPadminiRetina#1 73.9044747
## prdl.descr.my.fctriPadminiRetina#0 71.0522837
## prdl.descr.my.fctriPadAir#0 59.1826188
## prdl.descr.my.fctriPadmini3#0 55.8403584
## prdl.descr.my.fctriPadmini3#1 55.3587999
## prdl.descr.my.fctriPad5#0 54.8753087
## biddable 50.0303408
## prdl.descr.my.fctriPadmini2#0 48.3622392
## prdl.descr.my.fctriPad4#0 44.8353463
## prdl.descr.my.fctriPadAir#1 41.7019899
## prdl.descr.my.fctriPadmini2#1 38.7403007
## cellular.fctr1:carrier.fctrSprint 33.6915454
## prdl.descr.my.fctriPad3#0 32.7924348
## prdl.descr.my.fctriPad3#1 32.0088221
## cellular.fctr1:carrier.fctrT-Mobile 29.4854703
## condition.fctrNew other (see details) 28.3163808
## prdl.descr.my.fctrUnknown#1 26.7944874
## prdl.descr.my.fctriPadmini#0 26.2764947
## cellular.fctr1:carrier.fctrVerizon 25.7566816
## prdl.descr.my.fctriPad4#1 25.7337498
## prdl.descr.my.fctriPadmini#1 25.4865208
## condition.fctrNew 24.8929146
## prdl.descr.my.fctriPad2#0 24.8220248
## condition.fctrManufacturer refurbished 24.4120678
## cellular.fctr1:carrier.fctrUnknown 22.1568866
## .rnorm 22.0400171
## cellular.fctr1 22.0254370
## cellular.fctr0:carrier.fctrNone 20.8359173
## cellular.fctr0:carrier.fctrOther 20.8359173
## cellular.fctr0:carrier.fctrSprint 20.8359173
## cellular.fctr0:carrier.fctrT-Mobile 20.8359173
## cellular.fctr0:carrier.fctrUnknown 20.8359173
## cellular.fctr0:carrier.fctrVerizon 20.8359173
## cellular.fctr1:carrier.fctrNone 20.8359173
## cellular.fctrUnknown:carrier.fctrNone 20.8359173
## cellular.fctrUnknown:carrier.fctrOther 20.8359173
## cellular.fctrUnknown:carrier.fctrSprint 20.8359173
## cellular.fctrUnknown:carrier.fctrT-Mobile 20.8359173
## cellular.fctrUnknown:carrier.fctrVerizon 20.8359173
## prdl.descr.my.fctriPad2#1 20.7799709
## startprice 20.6165515
## color.fctrUnknown 18.4227302
## color.fctrSpace Gray 18.1370645
## color.fctrWhite 17.3387891
## prdl.descr.my.fctriPad1#1 16.2551004
## cellular.fctrUnknown 15.0866653
## cellular.fctrUnknown:carrier.fctrUnknown 14.5085844
## prdl.descr.my.fctriPad1#0 13.8044660
## storage.fctr64 10.9526155
## color.fctrGold 10.7220100
## condition.fctrFor parts or not working 9.9227633
## condition.fctrSeller refurbished 9.5602685
## storage.fctrUnknown 8.2731539
## storage.fctr32 0.8844774
## storage.fctr16 0.0000000if (glb_is_classification && glb_is_binomial)
glb_analytics_diag_plots(obs_df=glb_trnobs_df, mdl_id=glb_fin_mdl_id,
prob_threshold=glb_models_df[glb_models_df$id == glb_sel_mdl_id,
"opt.prob.threshold.OOB"]) else
glb_analytics_diag_plots(obs_df=glb_trnobs_df, mdl_id=glb_fin_mdl_id) ## Warning in glb_analytics_diag_plots(obs_df = glb_trnobs_df, mdl_id =
## glb_fin_mdl_id, : Limiting important feature scatter plots to 5 out of 9
## [1] "Min/Max Boundaries: "
## [1] UniqueID
## [2] sold.fctr
## [3] sold.fctr.predict.RFE.X.glmnet.prob
## [4] sold.fctr.predict.RFE.X.glmnet
## [5] sold.fctr.predict.RFE.X.glmnet.err
## [6] sold.fctr.predict.RFE.X.glmnet.err.abs
## [7] sold.fctr.predict.RFE.X.glmnet.accurate
## [8] sold.fctr.predict.Final.glmnet.prob
## [9] sold.fctr.predict.Final.glmnet
## [10] sold.fctr.predict.Final.glmnet.err
## [11] sold.fctr.predict.Final.glmnet.err.abs
## [12] sold.fctr.predict.Final.glmnet.accurate
## [13] sold.fctr.predict.Final.glmnet.error
## [14] .label
## <0 rows> (or 0-length row.names)
## [1] "Inaccurate: "
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 1704 11705 Y NA
## 1358 11359 Y 0.02691986
## 1816 11817 Y NA
## 1803 11804 Y NA
## 1523 11524 Y NA
## 525 10525 Y NA
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 1704 <NA> NA
## 1358 N FALSE
## 1816 <NA> NA
## 1803 <NA> NA
## 1523 <NA> NA
## 525 <NA> NA
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 1704 NA
## 1358 0.9730801
## 1816 NA
## 1803 NA
## 1523 NA
## 525 NA
## sold.fctr.predict.RFE.X.glmnet.accurate
## 1704 NA
## 1358 FALSE
## 1816 NA
## 1803 NA
## 1523 NA
## 525 NA
## sold.fctr.predict.Final.glmnet.prob sold.fctr.predict.Final.glmnet
## 1704 0.001477871 N
## 1358 0.002513302 N
## 1816 0.017219214 N
## 1803 0.019473215 N
## 1523 0.026338251 N
## 525 0.047825108 N
## sold.fctr.predict.Final.glmnet.err
## 1704 FALSE
## 1358 FALSE
## 1816 FALSE
## 1803 FALSE
## 1523 FALSE
## 525 FALSE
## sold.fctr.predict.Final.glmnet.err.abs
## 1704 0.9985221
## 1358 0.9974867
## 1816 0.9827808
## 1803 0.9805268
## 1523 0.9736617
## 525 0.9521749
## sold.fctr.predict.Final.glmnet.accurate
## 1704 FALSE
## 1358 FALSE
## 1816 FALSE
## 1803 FALSE
## 1523 FALSE
## 525 FALSE
## sold.fctr.predict.Final.glmnet.error
## 1704 -0.4985221
## 1358 -0.4974867
## 1816 -0.4827808
## 1803 -0.4805268
## 1523 -0.4736617
## 525 -0.4521749
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 353 10353 Y 0.1980042
## 21 10021 Y 0.3483605
## 1240 11241 Y NA
## 1458 11459 N NA
## 1092 11092 N NA
## 25 10025 N NA
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 353 N FALSE
## 21 N FALSE
## 1240 <NA> NA
## 1458 <NA> NA
## 1092 <NA> NA
## 25 <NA> NA
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 353 0.8019958
## 21 0.6516395
## 1240 NA
## 1458 NA
## 1092 NA
## 25 NA
## sold.fctr.predict.RFE.X.glmnet.accurate
## 353 FALSE
## 21 FALSE
## 1240 NA
## 1458 NA
## 1092 NA
## 25 NA
## sold.fctr.predict.Final.glmnet.prob sold.fctr.predict.Final.glmnet
## 353 0.1839583 N
## 21 0.3304191 N
## 1240 0.4865456 N
## 1458 0.5610813 Y
## 1092 0.6442881 Y
## 25 0.8688154 Y
## sold.fctr.predict.Final.glmnet.err
## 353 FALSE
## 21 FALSE
## 1240 FALSE
## 1458 FALSE
## 1092 FALSE
## 25 FALSE
## sold.fctr.predict.Final.glmnet.err.abs
## 353 0.8160417
## 21 0.6695809
## 1240 0.5134544
## 1458 0.5610813
## 1092 0.6442881
## 25 0.8688154
## sold.fctr.predict.Final.glmnet.accurate
## 353 FALSE
## 21 FALSE
## 1240 FALSE
## 1458 FALSE
## 1092 FALSE
## 25 FALSE
## sold.fctr.predict.Final.glmnet.error
## 353 -0.31604168
## 21 -0.16958091
## 1240 -0.01345439
## 1458 0.06108135
## 1092 0.14428810
## 25 0.36881540
## UniqueID sold.fctr sold.fctr.predict.RFE.X.glmnet.prob
## 491 10491 N NA
## 1249 11250 N NA
## 1390 11391 N NA
## 1505 11506 N 0.7978426
## 1470 11471 N NA
## 594 10594 N 0.6382435
## sold.fctr.predict.RFE.X.glmnet sold.fctr.predict.RFE.X.glmnet.err
## 491 <NA> NA
## 1249 <NA> NA
## 1390 <NA> NA
## 1505 Y FALSE
## 1470 <NA> NA
## 594 Y FALSE
## sold.fctr.predict.RFE.X.glmnet.err.abs
## 491 NA
## 1249 NA
## 1390 NA
## 1505 0.7978426
## 1470 NA
## 594 0.6382435
## sold.fctr.predict.RFE.X.glmnet.accurate
## 491 NA
## 1249 NA
## 1390 NA
## 1505 FALSE
## 1470 NA
## 594 FALSE
## sold.fctr.predict.Final.glmnet.prob sold.fctr.predict.Final.glmnet
## 491 0.9187431 Y
## 1249 0.9343948 Y
## 1390 0.9477133 Y
## 1505 0.9557130 Y
## 1470 0.9593088 Y
## 594 0.9752247 Y
## sold.fctr.predict.Final.glmnet.err
## 491 FALSE
## 1249 FALSE
## 1390 FALSE
## 1505 FALSE
## 1470 FALSE
## 594 FALSE
## sold.fctr.predict.Final.glmnet.err.abs
## 491 0.9187431
## 1249 0.9343948
## 1390 0.9477133
## 1505 0.9557130
## 1470 0.9593088
## 594 0.9752247
## sold.fctr.predict.Final.glmnet.accurate
## 491 FALSE
## 1249 FALSE
## 1390 FALSE
## 1505 FALSE
## 1470 FALSE
## 594 FALSE
## sold.fctr.predict.Final.glmnet.error
## 491 0.4187431
## 1249 0.4343948
## 1390 0.4477133
## 1505 0.4557130
## 1470 0.4593088
## 594 0.4752247
dsp_feats_vctr <- c(NULL)
for(var in grep(".importance", names(glb_feats_df), fixed=TRUE, value=TRUE))
dsp_feats_vctr <- union(dsp_feats_vctr,
glb_feats_df[!is.na(glb_feats_df[, var]), "id"])
# print(glb_trnobs_df[glb_trnobs_df$UniqueID %in% FN_OOB_ids,
# grep(glb_rsp_var, names(glb_trnobs_df), value=TRUE)])
print(setdiff(names(glb_trnobs_df), names(glb_allobs_df)))## [1] "sold.fctr.predict.Final.glmnet.prob"
## [2] "sold.fctr.predict.Final.glmnet"
## [3] "sold.fctr.predict.Final.glmnet.err"
## [4] "sold.fctr.predict.Final.glmnet.err.abs"
## [5] "sold.fctr.predict.Final.glmnet.accurate"for (col in setdiff(names(glb_trnobs_df), names(glb_allobs_df)))
# Merge or cbind ?
glb_allobs_df[glb_allobs_df$.src == "Train", col] <- glb_trnobs_df[, col]
print(setdiff(names(glb_fitobs_df), names(glb_allobs_df)))## character(0)print(setdiff(names(glb_OOBobs_df), names(glb_allobs_df)))## character(0)for (col in setdiff(names(glb_OOBobs_df), names(glb_allobs_df)))
# Merge or cbind ?
glb_allobs_df[glb_allobs_df$.lcn == "OOB", col] <- glb_OOBobs_df[, col]
print(setdiff(names(glb_newobs_df), names(glb_allobs_df)))## character(0)if (glb_save_envir)
save(glb_feats_df, glb_allobs_df,
#glb_trnobs_df, glb_fitobs_df, glb_OOBobs_df, glb_newobs_df,
glb_models_df, dsp_models_df, glb_models_lst, glb_model_type,
glb_sel_mdl, glb_sel_mdl_id,
glb_fin_mdl, glb_fin_mdl_id,
file=paste0(glb_out_pfx, "dsk.RData"))
replay.petrisim(pn=glb_analytics_pn,
replay.trans=(glb_analytics_avl_objs <- c(glb_analytics_avl_objs,
"data.training.all.prediction","model.final")), flip_coord=TRUE)## time trans "bgn " "fit.data.training.all " "predict.data.new " "end "
## 0.0000 multiple enabled transitions: data.training.all data.new model.selected firing: data.training.all
## 1.0000 1 2 1 0 0
## 1.0000 multiple enabled transitions: data.training.all data.new model.selected model.final data.training.all.prediction firing: data.new
## 2.0000 2 1 1 1 0
## 2.0000 multiple enabled transitions: data.training.all data.new model.selected model.final data.training.all.prediction data.new.prediction firing: model.selected
## 3.0000 3 0 2 1 0
## 3.0000 multiple enabled transitions: model.final data.training.all.prediction data.new.prediction firing: data.training.all.prediction
## 4.0000 5 0 1 1 1
## 4.0000 multiple enabled transitions: model.final data.training.all.prediction data.new.prediction firing: model.final
## 5.0000 4 0 0 2 1
glb_chunks_df <- myadd_chunk(glb_chunks_df, "predict.data.new", major.inc=TRUE)## label step_major step_minor label_minor bgn end
## 15 fit.data.training 8 1 1 223.102 230.388
## 16 predict.data.new 9 0 0 230.389 NA
## elapsed
## 15 7.286
## 16 NAStep 9.0: predict data new
# Compute final model predictions
#glb_to_sav(); all.equal(sav_allobs_df, glb_allobs_df); all.equal(sav_trnobs_df, glb_trnobs_df); all.equal(sav_newobs_df, glb_newobs_df)
if (glb_is_classification && glb_is_binomial)
prob_threshold_def <-
glb_models_df[glb_models_df$id == glb_sel_mdl_id, "opt.prob.threshold.OOB"] else
prob_threshold_def <- NULL
for (obsSet in c("trn", "new")) {
obs_df <- switch(obsSet, all = glb_allobs_df, trn = glb_trnobs_df, new = glb_newobs_df)
obs_df <- glb_get_predictions(obs_df, mdl_id = glb_fin_mdl_id,
rsp_var_out = glb_rsp_var_out, prob_threshold_def = prob_threshold_def)
if (obsSet == "all") glb_allobs_df <- obs_df else
if (obsSet == "trn") glb_trnobs_df <- obs_df else
if (obsSet == "new") glb_newobs_df <- obs_df
}## Warning in glb_get_predictions(obs_df, mdl_id = glb_fin_mdl_id, rsp_var_out
## = glb_rsp_var_out, : Using default probability threshold: 0.5## Warning in glb_get_predictions(obs_df, mdl_id = glb_fin_mdl_id, rsp_var_out
## = glb_rsp_var_out, : Using default probability threshold: 0.5rm(obs_df)
glb_allobs_df <- orderBy(reformulate(glb_id_var), myrbind_df(glb_trnobs_df, glb_newobs_df))
glb_analytics_diag_plots(obs_df = glb_newobs_df, mdl_id = glb_fin_mdl_id,
prob_threshold = prob_threshold_def)## Warning in glb_analytics_diag_plots(obs_df = glb_newobs_df, mdl_id =
## glb_fin_mdl_id, : Limiting important feature scatter plots to 5 out of 9## Warning: Removed 798 rows containing missing values (geom_point).## Warning: Removed 798 rows containing missing values (geom_point).
## Warning: Removed 798 rows containing missing values (geom_point).## Warning: Removed 798 rows containing missing values (geom_point).
## Warning: Removed 798 rows containing missing values (geom_point).## Warning: Removed 798 rows containing missing values (geom_point).
## Warning: Removed 798 rows containing missing values (geom_point).## Warning: Removed 798 rows containing missing values (geom_point).
## Warning: Removed 798 rows containing missing values (geom_point).## Warning: Removed 798 rows containing missing values (geom_point).
## [1] "Min/Max Boundaries: "
## [1] UniqueID
## [2] sold.fctr
## [3] sold.fctr.predict.Final.glmnet.prob
## [4] sold.fctr.predict.Final.glmnet
## [5] sold.fctr.predict.Final.glmnet.err
## [6] sold.fctr.predict.Final.glmnet.err.abs
## [7] sold.fctr.predict.Final.glmnet.accurate
## [8] sold.fctr.predict.Final.glmnet.error
## [9] .label
## <0 rows> (or 0-length row.names)
## [1] "Inaccurate: "
## UniqueID sold.fctr sold.fctr.predict.Final.glmnet.prob
## NA NA <NA> NA
## NA.1 NA <NA> NA
## NA.2 NA <NA> NA
## NA.3 NA <NA> NA
## NA.4 NA <NA> NA
## NA.5 NA <NA> NA
## sold.fctr.predict.Final.glmnet sold.fctr.predict.Final.glmnet.err
## NA <NA> NA
## NA.1 <NA> NA
## NA.2 <NA> NA
## NA.3 <NA> NA
## NA.4 <NA> NA
## NA.5 <NA> NA
## sold.fctr.predict.Final.glmnet.err.abs
## NA NA
## NA.1 NA
## NA.2 NA
## NA.3 NA
## NA.4 NA
## NA.5 NA
## sold.fctr.predict.Final.glmnet.accurate
## NA NA
## NA.1 NA
## NA.2 NA
## NA.3 NA
## NA.4 NA
## NA.5 NA
## sold.fctr.predict.Final.glmnet.error
## NA NA
## NA.1 NA
## NA.2 NA
## NA.3 NA
## NA.4 NA
## NA.5 NA
## UniqueID sold.fctr sold.fctr.predict.Final.glmnet.prob
## NA.80 NA <NA> NA
## NA.333 NA <NA> NA
## NA.433 NA <NA> NA
## NA.434 NA <NA> NA
## NA.460 NA <NA> NA
## NA.740 NA <NA> NA
## sold.fctr.predict.Final.glmnet sold.fctr.predict.Final.glmnet.err
## NA.80 <NA> NA
## NA.333 <NA> NA
## NA.433 <NA> NA
## NA.434 <NA> NA
## NA.460 <NA> NA
## NA.740 <NA> NA
## sold.fctr.predict.Final.glmnet.err.abs
## NA.80 NA
## NA.333 NA
## NA.433 NA
## NA.434 NA
## NA.460 NA
## NA.740 NA
## sold.fctr.predict.Final.glmnet.accurate
## NA.80 NA
## NA.333 NA
## NA.433 NA
## NA.434 NA
## NA.460 NA
## NA.740 NA
## sold.fctr.predict.Final.glmnet.error
## NA.80 NA
## NA.333 NA
## NA.433 NA
## NA.434 NA
## NA.460 NA
## NA.740 NA
## UniqueID sold.fctr sold.fctr.predict.Final.glmnet.prob
## NA.792 NA <NA> NA
## NA.793 NA <NA> NA
## NA.794 NA <NA> NA
## NA.795 NA <NA> NA
## NA.796 NA <NA> NA
## NA.797 NA <NA> NA
## sold.fctr.predict.Final.glmnet sold.fctr.predict.Final.glmnet.err
## NA.792 <NA> NA
## NA.793 <NA> NA
## NA.794 <NA> NA
## NA.795 <NA> NA
## NA.796 <NA> NA
## NA.797 <NA> NA
## sold.fctr.predict.Final.glmnet.err.abs
## NA.792 NA
## NA.793 NA
## NA.794 NA
## NA.795 NA
## NA.796 NA
## NA.797 NA
## sold.fctr.predict.Final.glmnet.accurate
## NA.792 NA
## NA.793 NA
## NA.794 NA
## NA.795 NA
## NA.796 NA
## NA.797 NA
## sold.fctr.predict.Final.glmnet.error
## NA.792 NA
## NA.793 NA
## NA.794 NA
## NA.795 NA
## NA.796 NA
## NA.797 NA## Warning: Removed 798 rows containing missing values (geom_point).
if (is.null(glb_out_obs)) obs_df <- glb_newobs_df else
obs_df <- switch(glb_out_obs,
all = glb_allobs_df, trn = glb_trnobs_df, new = glb_newobs_df)
require(stringr)## Loading required package: stringrif (glb_is_classification && glb_is_binomial) {
# submit_df <- glb_newobs_df[, c(glb_id_var,
# paste0(glb_rsp_var_out, glb_fin_mdl_id, ".prob"))]
# names(submit_df)[2] <- "Probability1"
obsout_df <- obs_df[, glb_id_var, FALSE]
for (clmn in names(glb_out_vars_lst))
if (!grepl("^%<d-%", glb_out_vars_lst[[clmn]]))
obsout_df[, clmn] <- obs_df[, glb_out_vars_lst[[clmn]]] else {
feat <- str_trim(unlist(strsplit(glb_out_vars_lst[[clmn]], "%<d-%"))[2])
obsout_df[, clmn] <- obs_df[, eval(parse(text = feat))]
}
glb_force_prediction_lst <- list()
glb_force_prediction_lst[["0"]] <- c(11885, 11907, 11932, 11943,
12050, 12115, 12171,
12253, 12285, 12367, 12388, 12399,
12585)
for (obs_id in glb_force_prediction_lst[["0"]]) {
if (is.na(glb_allobs_df[glb_allobs_df[, glb_id_var] == obs_id, ".grpid"]))
stop(".grpid is NA")
# submit_df[submit_df[, glb_id_var] == obs_id, "Probability1"] <-
# max(0, submit_df[submit_df[, glb_id_var] == obs_id, "Probability1"] - 0.5)
}
glb_force_prediction_lst[["1"]] <- c(11871, 11875, 11886,
11913, 11931, 11937, 11967, 11982, 11990, 11991, 11994, 11999,
12000, 12002, 12004, 12018, 12021, 12065, 12072,
12111, 12114, 12126, 12134, 12152, 12172,
12213, 12214, 12233, 12265, 12278, 12299,
12446, 12491,
12505, 12576, 12608, 12630)
for (obs_id in glb_force_prediction_lst[["1"]]) {
if (is.na(glb_allobs_df[glb_allobs_df[, glb_id_var] == obs_id, ".grpid"]))
stop(".grpid is NA")
# submit_df[submit_df[, glb_id_var] == obs_id, "Probability1"] <-
# min(0.9999, submit_df[submit_df[, glb_id_var] == obs_id, "Probability1"] + 0.5)
}
rsp_var_out <- paste0(glb_rsp_var_out, glb_fin_mdl_id)
for (obs_id in glb_newobs_df[!is.na(glb_newobs_df[, rsp_var_out]) &
(glb_newobs_df[, rsp_var_out] == "Y") &
(glb_newobs_df[ , "startprice"] > 675), "UniqueID"]) {
# submit_df[submit_df[, glb_id_var] == obs_id, "Probability1"] <-
# max(0, submit_df[submit_df[, glb_id_var] == obs_id, "Probability1"] - 0.5)
}
} else {
# submit_df <- glb_newobs_df[, c(glb_id_var,
# paste0(glb_rsp_var_out, glb_fin_mdl_id))]
obsout_df <- obs_df[, glb_id_var, FALSE]
for (clmn in names(glb_out_vars_lst))
if (!grepl("^%<d-%", glb_out_vars_lst[[clmn]]))
obsout_df[, clmn] <- obs_df[, glb_out_vars_lst[[clmn]]] else {
feat <- str_trim(unlist(strsplit(glb_out_vars_lst[[clmn]], "%<d-%"))[2])
obsout_df[, clmn] <- obs_df[, eval(parse(text=feat))]
}
}
if (glb_is_classification) {
rsp_var_out <- paste0(glb_rsp_var_out, glb_fin_mdl_id)
tmp_newobs_df <- subset(glb_newobs_df[, c(glb_id_var, ".grpid", rsp_var_out)],
!is.na(.grpid))
tmp_newobs_df <- merge(tmp_newobs_df, dupgrps_df, by = ".grpid", all.x = TRUE)
tmp_newobs_df <- merge(tmp_newobs_df, obsout_df, by = glb_id_var, all.x = TRUE)
tmp_newobs_df$.err <-
((tmp_newobs_df$Probability1 > 0.5) & (tmp_newobs_df$sold.0 > 0) |
(tmp_newobs_df$Probability1 < 0.5) & (tmp_newobs_df$sold.1 > 0))
tmp_newobs_df <- orderBy(~UniqueID, subset(tmp_newobs_df, .err == TRUE))
print("Prediction errors in duplicates:")
print(tmp_newobs_df)
# if (nrow(tmp_newobs_df) > 0)
# stop("check Prediction errors in duplicates")
#print(dupobs_df[dupobs_df$.grpid == 26, ])
tmp_newobs_df <- cbind(glb_newobs_df, obsout_df[, "Probability1", FALSE])
# if (max(glb_newobs_df[!is.na(glb_newobs_df[, rsp_var_out]) &
# (tmp_newobs_df[, "Probability1"] >= 0.5), "startprice"]) >
# max(glb_allobs_df[!is.na(glb_allobs_df[, glb_rsp_var]) &
# (glb_allobs_df[, glb_rsp_var] == "Y"), "startprice"]))
# stop("startprice for some +ve predictions > 675")
# Check predictions that are outside of data ranges
#stop(here")
require(stringr)
tmp_feats_df <- subset(glb_feats_df,
!nzv &
(exclude.as.feat != 1) &
!grepl(".fctr", id, fixed=TRUE))[, "id", FALSE]
ranges_all_df <- glb_allobs_df[, tmp_feats_df$id] %>%
dplyr::summarise_each(funs(min(., na.rm=TRUE),
max(., na.rm=TRUE))) %>%
tidyr::gather() %>%
dplyr::mutate(id=str_sub(key, 1, -5),
stat=str_sub(key, -3)) %>%
dplyr::select(-key) %>%
tidyr::spread(stat, value)
# sav_ranges_trn_df <- ranges_trn_df; all.equal(sav_ranges_trn_df, ranges_trn_df)
# sav_ranges_new_df <- ranges_new_df; all.equal(sav_ranges_new_df, ranges_new_df)
get_ranges_df <- function(obs_df, feats, class_var) {
require(tidyr)
ranges_df <- obs_df[, c(class_var, feats)] %>%
dplyr::group_by_(class_var) %>%
dplyr::summarise_each(funs(min(., na.rm=TRUE),
max(., na.rm=TRUE))) %>%
tidyr::gather(key, value, -1) %>%
mutate(id=str_sub(key, 1, -5),
stat.vname=paste0(str_sub(key, -3), ".", class_var)) %>%
unite_("stat.class", c("stat.vname", class_var), sep=".") %>%
dplyr::select(-key) %>%
spread(stat.class, value)
return(ranges_df)
}
rsp_var_out_OOB <- paste0(glb_rsp_var_out, glb_sel_mdl_id)
rsp_var_out_new <- paste0(glb_rsp_var_out, glb_fin_mdl_id)
ranges_trn_df <- get_ranges_df(obs_df=glb_trnobs_df, feats=tmp_feats_df$id,
class_var=glb_rsp_var)
ranges_fit_df <- get_ranges_df(obs_df=glb_fitobs_df, feats=tmp_feats_df$id,
class_var=glb_rsp_var)
ranges_OOB_df <- get_ranges_df(obs_df=glb_OOBobs_df, feats=tmp_feats_df$id,
class_var=rsp_var_out_OOB)
ranges_new_df <- get_ranges_df(obs_df=glb_newobs_df, feats=tmp_feats_df$id,
class_var=rsp_var_out_new)
for (obsset in c("OOB", "new")) {
if (obsset == "OOB") {
ranges_ref_df <- ranges_fit_df; obs_df <- glb_OOBobs_df;
rsp_var_out_obs <- rsp_var_out_OOB; sprintf_pfx <- "OOBobs";
} else {
ranges_ref_df <- ranges_trn_df; obs_df <- glb_newobs_df;
rsp_var_out_obs <- rsp_var_out_new; sprintf_pfx <- "newobs";
}
plt_feats_df <- glb_feats_df %>%
merge(ranges_all_df, all=TRUE) %>%
merge(ranges_ref_df, all=TRUE) %>%
merge(ranges_OOB_df, all=TRUE) %>%
merge(ranges_new_df, all=TRUE) %>%
subset(!is.na(min) & (id != ".rnorm"))
row.names(plt_feats_df) <- plt_feats_df$id
range_outlier_ids <- c(NULL)
for (clss in unique(obs_df[, rsp_var_out_obs])) {
for (stat in c("min", "max")) {
if (stat == "min") {
dsp_feats <- plt_feats_df[
which(plt_feats_df[, paste("min", rsp_var_out_obs, clss, sep=".")] <
plt_feats_df[, paste("min", glb_rsp_var, clss, sep=".")]), "id"]
} else {
dsp_feats <- plt_feats_df[
which(plt_feats_df[, paste("max", rsp_var_out_obs, clss, sep=".")] >
plt_feats_df[, paste("max", glb_rsp_var, clss, sep=".")]), "id"]
}
if (length(dsp_feats) > 0) {
ths_ids <- c(NULL)
for (feat in dsp_feats) {
if (stat == "min") {
ths_ids <- union(ths_ids,
obs_df[(obs_df[, rsp_var_out_obs] == clss) &
(obs_df[, feat] <
plt_feats_df[plt_feats_df$id == feat, paste("min", glb_rsp_var, clss, sep=".")]),
glb_id_var])
} else {
ths_ids <- union(ths_ids,
obs_df[(obs_df[, rsp_var_out_obs] == clss) &
(obs_df[, feat] >
plt_feats_df[plt_feats_df$id == feat, paste("max", glb_rsp_var, clss, sep=".")]),
glb_id_var])
}
}
tmp_obs_df <- obs_df[obs_df[, glb_id_var] %in% ths_ids,
c(glb_id_var, rsp_var_out_obs, dsp_feats)]
if (stat == "min") {
print(sprintf("%s %s %s: min < min of Train range: %d",
sprintf_pfx, rsp_var_out_obs, clss, nrow(tmp_obs_df)))
} else {
print(sprintf("%s %s %s: max > max of Train range: %d",
sprintf_pfx, rsp_var_out_obs, clss, nrow(tmp_obs_df)))
}
myprint_df(tmp_obs_df)
print(subset(plt_feats_df, id %in% dsp_feats))
range_outlier_ids <- union(range_outlier_ids, ths_ids)
}
}
}
print(sprintf("%s total range outliers: %d", sprintf_pfx, length(range_outlier_ids)))
}
} ## [1] "Prediction errors in duplicates:"
## UniqueID .grpid sold.fctr.predict.Final.glmnet sold.0 sold.1 sold.NA
## 4 11886 134 N 0 1 1
## 9 11931 53 N 0 2 2
## 23 11994 101 N 0 1 1
## 24 11999 115 N 0 1 1
## 44 12115 56 Y 1 0 1
## 59 12233 30 N 0 1 1
## 63 12253 57 Y 1 0 1
## 68 12285 55 Y 1 0 2
## 69 12299 53 N 0 2 2
## 72 12367 55 Y 1 0 2
## 73 12388 124 Y 1 0 1
## 76 12446 103 N 0 1 1
## 80 12505 54 N 0 1 1
## 81 12576 93 N 0 1 1
## 82 12585 112 Y 1 0 1
## .freq Probability1 .err
## 4 2 0.1590218 TRUE
## 9 4 0.2836153 TRUE
## 23 2 0.2998070 TRUE
## 24 2 0.3858952 TRUE
## 44 2 0.5143733 TRUE
## 59 2 0.2903443 TRUE
## 63 2 0.6319957 TRUE
## 68 3 0.6135146 TRUE
## 69 4 0.2759853 TRUE
## 72 3 0.6131161 TRUE
## 73 2 0.6167362 TRUE
## 76 2 0.2358354 TRUE
## 80 2 0.3128248 TRUE
## 81 2 0.2464955 TRUE
## 82 2 0.6264435 TRUE## Loading required package: tidyr
##
## Attaching package: 'tidyr'
##
## The following object is masked from 'package:Matrix':
##
## expand## [1] "OOBobs sold.fctr.predict.RFE.X.glmnet N: min < min of Train range: 1"
## UniqueID sold.fctr.predict.RFE.X.glmnet startprice
## 1590 11591 N 2.99
## id cor.y exclude.as.feat cor.y.abs cor.high.X
## startprice startprice -0.4569767 FALSE 0.4569767 <NA>
## freqRatio percentUnique zeroVar nzv is.cor.y.abs.low
## startprice 2.807692 30.17751 FALSE FALSE FALSE
## interaction.feat shapiro.test.p.value rsp_var_raw id_var
## startprice <NA> 1.656423e-29 FALSE NA
## rsp_var max min max.sold.fctr.N max.sold.fctr.Y
## startprice NA 999.99 0.01 939 650
## min.sold.fctr.N min.sold.fctr.Y
## startprice 3.99 0.01
## max.sold.fctr.predict.RFE.X.glmnet.N
## startprice 999
## max.sold.fctr.predict.RFE.X.glmnet.Y
## startprice 450
## min.sold.fctr.predict.RFE.X.glmnet.N
## startprice 2.99
## min.sold.fctr.predict.RFE.X.glmnet.Y
## startprice 0.01
## max.sold.fctr.predict.Final.glmnet.N
## startprice 999.99
## max.sold.fctr.predict.Final.glmnet.Y
## startprice 550
## min.sold.fctr.predict.Final.glmnet.N
## startprice 30
## min.sold.fctr.predict.Final.glmnet.Y
## startprice 0.01
## [1] "OOBobs sold.fctr.predict.RFE.X.glmnet N: max > max of Train range: 2"
## UniqueID sold.fctr.predict.RFE.X.glmnet startprice
## 1396 11397 N 999.00
## 1282 11283 N 948.98
## id cor.y exclude.as.feat cor.y.abs cor.high.X
## startprice startprice -0.4569767 FALSE 0.4569767 <NA>
## freqRatio percentUnique zeroVar nzv is.cor.y.abs.low
## startprice 2.807692 30.17751 FALSE FALSE FALSE
## interaction.feat shapiro.test.p.value rsp_var_raw id_var
## startprice <NA> 1.656423e-29 FALSE NA
## rsp_var max min max.sold.fctr.N max.sold.fctr.Y
## startprice NA 999.99 0.01 939 650
## min.sold.fctr.N min.sold.fctr.Y
## startprice 3.99 0.01
## max.sold.fctr.predict.RFE.X.glmnet.N
## startprice 999
## max.sold.fctr.predict.RFE.X.glmnet.Y
## startprice 450
## min.sold.fctr.predict.RFE.X.glmnet.N
## startprice 2.99
## min.sold.fctr.predict.RFE.X.glmnet.Y
## startprice 0.01
## max.sold.fctr.predict.Final.glmnet.N
## startprice 999.99
## max.sold.fctr.predict.Final.glmnet.Y
## startprice 550
## min.sold.fctr.predict.Final.glmnet.N
## startprice 30
## min.sold.fctr.predict.Final.glmnet.Y
## startprice 0.01
## [1] "OOBobs total range outliers: 3"
## [1] "newobs sold.fctr.predict.Final.glmnet N: max > max of Train range: 1"
## UniqueID sold.fctr.predict.Final.glmnet startprice
## 2623 12625 N 999.99
## id cor.y exclude.as.feat cor.y.abs cor.high.X
## startprice startprice -0.4569767 FALSE 0.4569767 <NA>
## freqRatio percentUnique zeroVar nzv is.cor.y.abs.low
## startprice 2.807692 30.17751 FALSE FALSE FALSE
## interaction.feat shapiro.test.p.value rsp_var_raw id_var
## startprice <NA> 1.656423e-29 FALSE NA
## rsp_var max min max.sold.fctr.N max.sold.fctr.Y
## startprice NA 999.99 0.01 999 675
## min.sold.fctr.N min.sold.fctr.Y
## startprice 0.99 0.01
## max.sold.fctr.predict.RFE.X.glmnet.N
## startprice 999
## max.sold.fctr.predict.RFE.X.glmnet.Y
## startprice 450
## min.sold.fctr.predict.RFE.X.glmnet.N
## startprice 2.99
## min.sold.fctr.predict.RFE.X.glmnet.Y
## startprice 0.01
## max.sold.fctr.predict.Final.glmnet.N
## startprice 999.99
## max.sold.fctr.predict.Final.glmnet.Y
## startprice 550
## min.sold.fctr.predict.Final.glmnet.N
## startprice 30
## min.sold.fctr.predict.Final.glmnet.Y
## startprice 0.01
## [1] "newobs total range outliers: 1"out_fname <- paste0(glb_out_pfx, "out.csv")
write.csv(obsout_df, out_fname, quote=FALSE, row.names=FALSE)
#cat(" ", "\n", file=submit_fn, append=TRUE)
# print(orderBy(~ -max.auc.OOB, glb_models_df[, c("id",
# "max.auc.OOB", "max.Accuracy.OOB")]))
for (txt_var in glb_txt_vars) {
# Print post-stem-words but need post-stop-words for debugging ?
print(sprintf(" All post-stem-words term weights for %s:", txt_var))
myprint_df(glb_post_stem_words_terms_df_lst[[txt_var]])
terms_mtrx <- glb_post_stem_words_terms_mtrx_lst[[txt_var]]
print(glb_allobs_df[
which(terms_mtrx[, tail(glb_post_stem_words_terms_df_lst[[txt_var]], 1)$pos] > 0),
c(glb_id_var, glb_txt_vars)])
print(nrow(subset(glb_post_stem_words_terms_df_lst[[txt_var]], freq == 1)))
#print(glb_allobs_df[which(terms_mtrx[, 207] > 0), c(glb_id_var, glb_txt_vars)])
#unlist(strsplit(glb_allobs_df[2157, "description"], ""))
#glb_allobs_df[2442, c(glb_id_var, glb_txt_vars)]
#terms_mtrx[2442, terms_mtrx[2442, ] > 0]
print(sprintf(" All post-stem-words term freq distribution for %s:", txt_var))
print(table(glb_post_stem_words_terms_df_lst[[txt_var]]$freq))
print(sprintf(" All post-stem-words term length distribution for %s:", txt_var))
print(table(nchar(glb_post_stem_words_terms_df_lst[[txt_var]]$term)))
print(subset(glb_post_stem_words_terms_df_lst[[txt_var]], nchar(term) >= 10))
print(sprintf(" Analyzed term weights for %s:", txt_var))
tmp_df <- glb_post_stem_words_terms_df_lst[[txt_var]]
anl_terms_vctr <- union(select_terms, assoc_terms)
print(subset(tmp_df, term %in% anl_terms_vctr))
# tmp_freq1_df <- subset(tmp_df, freq == 1)
# tmp_freq1_df$top_n <- grepl(paste0(top_n_vctr, collapse="|"), tmp_freq1_df$term)
# print(subset(tmp_freq1_df, top_n == TRUE))
}
if (glb_is_classification && glb_is_binomial)
print(glb_models_df[glb_models_df$id == glb_sel_mdl_id,
"opt.prob.threshold.OOB"])## [1] 0.5print(sprintf("glb_sel_mdl_id: %s", glb_sel_mdl_id))## [1] "glb_sel_mdl_id: RFE.X.glmnet"print(sprintf("glb_fin_mdl_id: %s", glb_fin_mdl_id))## [1] "glb_fin_mdl_id: Final.glmnet"get_dsp_models_df()## [1] "Cross Validation issues:"## Warning in get_dsp_models_df(): Cross Validation issues:## MFO.myMFO_classfr Random.myrandom_classfr
## 0 0
## Max.cor.Y.rcv.1X1.glmnet Max.cor.Y.rcv.1X1.cp.0.rpart
## 0 0## id
## Max.cor.Y.rcv.3X1.glmnet Max.cor.Y.rcv.3X1.glmnet
## Max.cor.Y.rcv.5X1.glmnet Max.cor.Y.rcv.5X1.glmnet
## Interact.High.cor.Y.glmnet Interact.High.cor.Y.glmnet
## Max.cor.Y.rcv.3X1.Interact.glmnet Max.cor.Y.rcv.3X1.Interact.glmnet
## Max.cor.Y.rcv.3X3.glmnet Max.cor.Y.rcv.3X3.glmnet
## Max.cor.Y.rcv.3X5.glmnet Max.cor.Y.rcv.3X5.glmnet
## Max.cor.Y.rcv.5X3.glmnet Max.cor.Y.rcv.5X3.glmnet
## Max.cor.Y.rcv.5X5.glmnet Max.cor.Y.rcv.5X5.glmnet
## Max.cor.Y.rcv.1X1.glmnet Max.cor.Y.rcv.1X1.glmnet
## RFE.X.rf RFE.X.rf
## RFE.X.gbm RFE.X.gbm
## RFE.X.glmnet RFE.X.glmnet
## All.X.glmnet All.X.glmnet
## Low.cor.X.glmnet Low.cor.X.glmnet
## RFE.X.bayesglm RFE.X.bayesglm
## Max.cor.Y.rpart Max.cor.Y.rpart
## RFE.X.rpart RFE.X.rpart
## RFE.X.glm RFE.X.glm
## Max.cor.Y.rcv.1X1.cp.0.rpart Max.cor.Y.rcv.1X1.cp.0.rpart
## Random.myrandom_classfr Random.myrandom_classfr
## MFO.myMFO_classfr MFO.myMFO_classfr
## Final.glm Final.glm
## Final.glmnet Final.glmnet
## max.auc.OOB max.Accuracy.OOB
## Max.cor.Y.rcv.3X1.glmnet 0.8266524 0.7844444
## Max.cor.Y.rcv.5X1.glmnet 0.8266524 0.7844444
## Interact.High.cor.Y.glmnet 0.8262948 0.7811111
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.8262948 0.7811111
## Max.cor.Y.rcv.3X3.glmnet 0.8260961 0.7811111
## Max.cor.Y.rcv.3X5.glmnet 0.8260961 0.7811111
## Max.cor.Y.rcv.5X3.glmnet 0.8260961 0.7811111
## Max.cor.Y.rcv.5X5.glmnet 0.8260961 0.7811111
## Max.cor.Y.rcv.1X1.glmnet 0.8253362 0.7777778
## RFE.X.rf 0.8240549 0.7366667
## RFE.X.gbm 0.8223538 0.7466667
## RFE.X.glmnet 0.8178863 0.7800000
## All.X.glmnet 0.8178863 0.7800000
## Low.cor.X.glmnet 0.8131655 0.7655556
## RFE.X.bayesglm 0.8111689 0.7655556
## Max.cor.Y.rpart 0.8095598 0.7777778
## RFE.X.rpart 0.8092816 0.7777778
## RFE.X.glm 0.8055666 0.7611111
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.7971655 0.7200000
## Random.myrandom_classfr 0.5162111 0.4622222
## MFO.myMFO_classfr 0.5000000 0.5377778
## Final.glm NA NA
## Final.glmnet NA NA
## max.Kappa.OOB opt.prob.threshold.OOB
## Max.cor.Y.rcv.3X1.glmnet 0.5639099 0.5
## Max.cor.Y.rcv.5X1.glmnet 0.5639099 0.5
## Interact.High.cor.Y.glmnet 0.5569405 0.5
## Max.cor.Y.rcv.3X1.Interact.glmnet 0.5569405 0.5
## Max.cor.Y.rcv.3X3.glmnet 0.5569405 0.5
## Max.cor.Y.rcv.3X5.glmnet 0.5569405 0.5
## Max.cor.Y.rcv.5X3.glmnet 0.5569405 0.5
## Max.cor.Y.rcv.5X5.glmnet 0.5569405 0.5
## Max.cor.Y.rcv.1X1.glmnet 0.5499640 0.5
## RFE.X.rf 0.4744804 0.4
## RFE.X.gbm 0.4917974 0.4
## RFE.X.glmnet 0.5562749 0.5
## All.X.glmnet 0.5562749 0.5
## Low.cor.X.glmnet 0.5277012 0.5
## RFE.X.bayesglm 0.5278609 0.5
## Max.cor.Y.rpart 0.5520962 0.3
## RFE.X.rpart 0.5520962 0.3
## RFE.X.glm 0.5192355 0.5
## Max.cor.Y.rcv.1X1.cp.0.rpart 0.4450317 0.3
## Random.myrandom_classfr 0.0000000 0.4
## MFO.myMFO_classfr 0.0000000 0.5
## Final.glm NA NA
## Final.glmnet NA NAif (glb_is_regression) {
print(sprintf("%s OOB RMSE: %0.4f", glb_sel_mdl_id,
glb_models_df[glb_models_df$id == glb_sel_mdl_id, "min.RMSE.OOB"]))
if (!is.null(glb_category_var)) {
#stop(here"); glb_to_sav(); glb_ctgry_df <- sav_ctgry_df
# OOB_ctgry_df <- myget_category_stats(glb_OOBobs_df, glb_sel_mdl_id, "OOB")
# glb_ctgry_df <- merge(glb_ctgry_df, subset(OOB_ctgry_df, select=-.n.OOB),
# by=glb_category_var, all=TRUE)
#
# #glb_fitobs_df <- glb_get_predictions(glb_fitobs_df, glb_sel_mdl_id, glb_rsp_var_out)
# glb_ctgry_df <- merge(glb_ctgry_df,
# myget_category_stats(obs_df=glb_fitobs_df, mdl_id=glb_sel_mdl_id, label="fit"),
# by=glb_category_var, all=TRUE)
# row.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
glb_ctgry_df <- merge(glb_ctgry_df,
myget_category_stats(obs_df=glb_trnobs_df, mdl_id=glb_fin_mdl_id, label="trn"),
by=glb_category_var, all=TRUE)
row.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
glb_ctgry_df <- merge(glb_ctgry_df,
myget_category_stats(obs_df=glb_newobs_df, mdl_id=glb_fin_mdl_id, label="new"),
by=glb_category_var, all=TRUE)
row.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
if (any(grepl("OOB", glb_model_evl_criteria)))
print(orderBy(~-err.abs.OOB.mean, glb_ctgry_df)) else
print(orderBy(~-err.abs.fit.mean, glb_ctgry_df))
print(colSums(glb_ctgry_df[, -grep(glb_category_var, names(glb_ctgry_df))]))
}
if ((glb_rsp_var %in% names(glb_newobs_df)) &&
!(any(is.na(glb_newobs_df[, glb_rsp_var])))) {
pred_stats_df <-
mypredict_mdl(mdl=glb_models_lst[[glb_fin_mdl_id]],
df=glb_newobs_df,
rsp_var=glb_rsp_var,
rsp_var_out=glb_rsp_var_out,
mdl_id=glb_fin_mdl_id,
label="new",
model_summaryFunction=glb_sel_mdl$control$summaryFunction,
model_metric=glb_sel_mdl$metric,
model_metric_maximize=glb_sel_mdl$maximize,
ret_type="stats")
print(sprintf("%s prediction stats for glb_newobs_df:", glb_fin_mdl_id))
print(pred_stats_df)
}
}
if (glb_is_classification) {
print(sprintf("%s OOB confusion matrix & accuracy: ", glb_sel_mdl_id))
print(t(confusionMatrix(glb_OOBobs_df[, paste0(glb_rsp_var_out, glb_sel_mdl_id)],
glb_OOBobs_df[, glb_rsp_var])$table))
if (!is.null(glb_category_var)) {
glb_ctgry_df <- merge(glb_ctgry_df,
myget_category_stats(obs_df=glb_trnobs_df, mdl_id=glb_fin_mdl_id, label="trn"),
by=glb_category_var, all=TRUE)
row.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
glb_ctgry_df <- merge(glb_ctgry_df,
myget_category_stats(obs_df=glb_newobs_df, mdl_id=glb_fin_mdl_id, label="new"),
by=glb_category_var, all=TRUE)
row.names(glb_ctgry_df) <- glb_ctgry_df[, glb_category_var]
if (any(grepl("OOB", glb_model_evl_criteria)))
print(orderBy(~-err.abs.OOB.mean, glb_ctgry_df)) else
print(orderBy(~-err.abs.fit.mean, glb_ctgry_df))
print(colSums(glb_ctgry_df[, -grep(glb_category_var, names(glb_ctgry_df))]))
# print("Top category OOB errors:")
# print(glb_OOBobs_df[(glb_OOBobs_df[, glb_category_var] ==
# dsp_ctgry_df[1, glb_category_var]) &
# !glb_OOBobs_df[, predct_accurate_var_name],
# c(glb_id_var, glb_rsp_var_raw, paste0(glb_rsp_var_out, glb_sel_mdl_id),
# glb_category_var,
# row.names(head(myget_feats_importance(glb_sel_mdl), 5)),
# # "biddable", "startprice", "condition",
# glb_txt_vars)])
}
if ((glb_rsp_var %in% names(glb_newobs_df)) &&
!(any(is.na(glb_newobs_df[, glb_rsp_var])))) {
print(sprintf("%s new confusion matrix & accuracy: ", glb_fin_mdl_id))
print(t(confusionMatrix(glb_newobs_df[, paste0(glb_rsp_var_out, glb_fin_mdl_id)],
glb_newobs_df[, glb_rsp_var])$table))
}
} ## [1] "RFE.X.glmnet OOB confusion matrix & accuracy: "
## Prediction
## Reference N Y
## N 393 91
## Y 107 309
## prdl.descr.my.fctr .n.Tst .n.OOB.x .freqRatio.Tst
## iPadminiRetina#1 iPadminiRetina#1 NA 2 NA
## iPad5#0 iPad5#0 NA 1 NA
## iPad2#0 iPad2#0 83 93 0.10401003
## iPad1#1 iPad1#1 42 47 0.05263158
## Unknown#1 Unknown#1 47 52 0.05889724
## iPad2#1 iPad2#1 71 79 0.08897243
## iPad1#0 iPad1#0 46 52 0.05764411
## iPadmini#1 iPadmini#1 46 52 0.05764411
## iPad3#0 iPad3#0 30 34 0.03759398
## iPadmini#0 iPadmini#0 65 73 0.08145363
## Unknown#0 Unknown#0 45 50 0.05639098
## iPad4#0 iPad4#0 29 33 0.03634085
## iPadmini3#0 iPadmini3#0 29 33 0.03634085
## iPadmini2#1 iPadmini2#1 21 24 0.02631579
## iPad3#1 iPad3#1 25 28 0.03132832
## iPadAir2#0 iPadAir2#0 47 52 0.05889724
## iPadAir2#1 iPadAir2#1 15 17 0.01879699
## iPadAir#0 iPadAir#0 41 46 0.05137845
## iPad4#1 iPad4#1 39 44 0.04887218
## iPadmini2#0 iPadmini2#0 35 39 0.04385965
## iPadmini3#1 iPadmini3#1 9 10 0.01127820
## iPadminiRetina#0 iPadminiRetina#0 NA 2 NA
## iPadAir#1 iPadAir#1 33 37 0.04135338
## .freqRatio.OOB err.abs.fit.sum err.abs.fit.mean .n.fit
## iPadminiRetina#1 0.002222222 NA NA NA
## iPad5#0 0.001111111 NA NA NA
## iPad2#0 0.103333333 9.3686815 0.2129246 44
## iPad1#1 0.052222222 19.9350980 0.3559839 56
## Unknown#1 0.057777778 12.1033794 0.3667691 33
## iPad2#1 0.087777778 22.2690424 0.3181292 70
## iPad1#0 0.057777778 23.4721073 0.3353158 70
## iPadmini#1 0.057777778 17.2037432 0.3308412 52
## iPad3#0 0.037777778 9.4308076 0.2548867 37
## iPadmini#0 0.081111111 34.6998466 0.3469985 100
## Unknown#0 0.055555556 26.1411829 0.3788577 69
## iPad4#0 0.036666667 13.1137754 0.3278444 40
## iPadmini3#0 0.036666667 8.9633487 0.1991855 45
## iPadmini2#1 0.026666667 4.6215934 0.3555072 13
## iPad3#1 0.031111111 18.4753679 0.3421364 54
## iPadAir2#0 0.057777778 20.9263375 0.2790178 75
## iPadAir2#1 0.018888889 6.6020765 0.2445214 27
## iPadAir#0 0.051111111 13.3454357 0.2839454 47
## iPad4#1 0.048888889 7.8246254 0.1956156 40
## iPadmini2#0 0.043333333 12.1895562 0.3932115 31
## iPadmini3#1 0.011111111 0.2077954 0.1038977 2
## iPadminiRetina#0 0.002222222 2.1227793 0.5306948 4
## iPadAir#1 0.041111111 11.9615695 0.2392314 50
## err.abs.OOB.sum err.abs.OOB.mean .n.OOB.y err.abs.trn.sum
## iPadminiRetina#1 0.9496705 0.4748353 2 0.771011349
## iPad5#0 0.4579943 0.4579943 1 0.006209032
## iPad2#0 35.8776572 0.3857813 93 40.278045532
## iPad1#1 17.2801034 0.3676618 47 34.600991046
## Unknown#1 19.0108650 0.3655936 52 28.932180025
## iPad2#1 28.7900280 0.3644307 79 45.288476577
## iPad1#0 18.5486352 0.3567045 52 38.338025581
## iPadmini#1 18.2005569 0.3500107 52 30.333049850
## iPad3#0 11.6504459 0.3426602 34 18.602080969
## iPadmini#0 24.6734074 0.3379919 73 52.449666294
## Unknown#0 16.7861405 0.3357228 50 41.343101737
## iPad4#0 11.0562541 0.3350380 33 20.842614700
## iPadmini3#0 10.9791835 0.3327025 33 17.953337229
## iPadmini2#1 7.9783124 0.3324297 24 11.204272576
## iPad3#1 9.1650063 0.3273217 28 23.938639227
## iPadAir2#0 16.6505538 0.3202030 52 34.047727948
## iPadAir2#1 5.2910291 0.3112370 17 10.277586994
## iPadAir#0 14.3134845 0.3111627 46 24.439356271
## iPad4#1 13.4125718 0.3048312 44 18.223908377
## iPadmini2#0 11.7336648 0.3008632 39 22.368417557
## iPadmini3#1 2.7265614 0.2726561 10 2.457434929
## iPadminiRetina#0 0.5069519 0.2534760 2 2.790257070
## iPadAir#1 9.2942141 0.2511950 37 17.432762337
## err.abs.trn.mean .n.trn err.abs.new.sum err.abs.new.mean
## iPadminiRetina#1 0.385505675 2 NA NA
## iPad5#0 0.006209032 1 NA NA
## iPad2#0 0.294000332 137 NA NA
## iPad1#1 0.335931952 103 NA NA
## Unknown#1 0.340378589 85 NA NA
## iPad2#1 0.303949507 149 NA NA
## iPad1#0 0.314246111 122 NA NA
## iPadmini#1 0.291663941 104 NA NA
## iPad3#0 0.262001140 71 NA NA
## iPadmini#0 0.303177262 173 NA NA
## Unknown#0 0.347421023 119 NA NA
## iPad4#0 0.285515270 73 NA NA
## iPadmini3#0 0.230170990 78 NA NA
## iPadmini2#1 0.302818178 37 NA NA
## iPad3#1 0.291934625 82 NA NA
## iPadAir2#0 0.268092346 127 NA NA
## iPadAir2#1 0.233581523 44 NA NA
## iPadAir#0 0.262788777 93 NA NA
## iPad4#1 0.216951290 84 NA NA
## iPadmini2#0 0.319548822 70 NA NA
## iPadmini3#1 0.204786244 12 NA NA
## iPadminiRetina#0 0.465042845 6 NA NA
## iPadAir#1 0.200376579 87 NA NA
## .n.new
## iPadminiRetina#1 NA
## iPad5#0 NA
## iPad2#0 83
## iPad1#1 42
## Unknown#1 47
## iPad2#1 71
## iPad1#0 46
## iPadmini#1 46
## iPad3#0 30
## iPadmini#0 65
## Unknown#0 45
## iPad4#0 29
## iPadmini3#0 29
## iPadmini2#1 21
## iPad3#1 25
## iPadAir2#0 47
## iPadAir2#1 15
## iPadAir#0 41
## iPad4#1 39
## iPadmini2#0 35
## iPadmini3#1 9
## iPadminiRetina#0 NA
## iPadAir#1 33
## .n.Tst .n.OOB.x .freqRatio.Tst .freqRatio.OOB
## NA 900.000000 NA 1.000000
## err.abs.fit.sum err.abs.fit.mean .n.fit err.abs.OOB.sum
## NA NA NA 305.333292
## err.abs.OOB.mean .n.OOB.y err.abs.trn.sum err.abs.trn.mean
## 7.792503 900.000000 536.919153 6.466092
## .n.trn err.abs.new.sum err.abs.new.mean .n.new
## 1859.000000 NA NA NAif (!is.null(glb_featsimp_df))
print(orderBy(as.formula(paste0("~ -", glb_sel_mdl_id, ".importance")),
subset(glb_featsimp_df, importance > 10)))## RFE.X.glmnet.importance
## biddable 100.00000
## prdl.descr.my.fctriPadminiRetina#0 69.85328
## prdl.descr.my.fctriPad2#0 63.75435
## prdl.descr.my.fctriPadAir2#0 62.34763
## prdl.descr.my.fctriPadAir#0 55.88955
## prdl.descr.my.fctriPadAir2#1 55.66092
## cellular.fctr1:carrier.fctrT-Mobile 50.56798
## prdl.descr.my.fctriPad4#0 50.37100
## prdl.descr.my.fctriPad3#0 47.81549
## prdl.descr.my.fctriPadmini2#0 44.08210
## cellular.fctr1 43.54682
## storage.fctr64 41.91171
## cellular.fctr1:carrier.fctrVerizon 41.54999
## cellular.fctr1:carrier.fctrUnknown 39.72591
## cellular.fctr1:carrier.fctrOther 38.77602
## prdl.descr.my.fctriPadminiRetina#1 38.77602
## prdl.descr.my.fctriPadmini3#1 38.77602
## prdl.descr.my.fctriPad5#0 38.77602
## prdl.descr.my.fctriPadAir#1 38.77602
## prdl.descr.my.fctriPadmini2#1 38.77602
## cellular.fctr1:carrier.fctrSprint 38.77602
## prdl.descr.my.fctriPad3#1 38.77602
## condition.fctrNew other (see details) 38.77602
## prdl.descr.my.fctriPadmini#1 38.77602
## condition.fctrNew 38.77602
## condition.fctrManufacturer refurbished 38.77602
## .rnorm 38.77602
## cellular.fctr0:carrier.fctrNone 38.77602
## cellular.fctr0:carrier.fctrOther 38.77602
## cellular.fctr0:carrier.fctrSprint 38.77602
## cellular.fctr0:carrier.fctrT-Mobile 38.77602
## cellular.fctr0:carrier.fctrUnknown 38.77602
## cellular.fctr0:carrier.fctrVerizon 38.77602
## cellular.fctr1:carrier.fctrNone 38.77602
## cellular.fctrUnknown:carrier.fctrNone 38.77602
## cellular.fctrUnknown:carrier.fctrOther 38.77602
## cellular.fctrUnknown:carrier.fctrSprint 38.77602
## cellular.fctrUnknown:carrier.fctrT-Mobile 38.77602
## cellular.fctrUnknown:carrier.fctrVerizon 38.77602
## color.fctrUnknown 38.77602
## color.fctrSpace Gray 38.77602
## color.fctrWhite 38.77602
## prdl.descr.my.fctriPad1#1 38.77602
## prdl.descr.my.fctriPad1#0 38.77602
## color.fctrGold 38.77602
## startprice 38.55010
## prdl.descr.my.fctrUnknown#1 38.11731
## prdl.descr.my.fctriPadmini3#0 36.37694
## prdl.descr.my.fctriPadmini#0 36.09147
## prdl.descr.my.fctriPad2#1 32.06133
## cellular.fctrUnknown:carrier.fctrUnknown 29.05086
## cellular.fctrUnknown 28.65476
## prdl.descr.my.fctriPad4#1 0.00000
## importance
## biddable 50.03034
## prdl.descr.my.fctriPadminiRetina#0 71.05228
## prdl.descr.my.fctriPad2#0 24.82202
## prdl.descr.my.fctriPadAir2#0 81.59513
## prdl.descr.my.fctriPadAir#0 59.18262
## prdl.descr.my.fctriPadAir2#1 73.99464
## cellular.fctr1:carrier.fctrT-Mobile 29.48547
## prdl.descr.my.fctriPad4#0 44.83535
## prdl.descr.my.fctriPad3#0 32.79243
## prdl.descr.my.fctriPadmini2#0 48.36224
## cellular.fctr1 22.02544
## storage.fctr64 10.95262
## cellular.fctr1:carrier.fctrVerizon 25.75668
## cellular.fctr1:carrier.fctrUnknown 22.15689
## cellular.fctr1:carrier.fctrOther 100.00000
## prdl.descr.my.fctriPadminiRetina#1 73.90447
## prdl.descr.my.fctriPadmini3#1 55.35880
## prdl.descr.my.fctriPad5#0 54.87531
## prdl.descr.my.fctriPadAir#1 41.70199
## prdl.descr.my.fctriPadmini2#1 38.74030
## cellular.fctr1:carrier.fctrSprint 33.69155
## prdl.descr.my.fctriPad3#1 32.00882
## condition.fctrNew other (see details) 28.31638
## prdl.descr.my.fctriPadmini#1 25.48652
## condition.fctrNew 24.89291
## condition.fctrManufacturer refurbished 24.41207
## .rnorm 22.04002
## cellular.fctr0:carrier.fctrNone 20.83592
## cellular.fctr0:carrier.fctrOther 20.83592
## cellular.fctr0:carrier.fctrSprint 20.83592
## cellular.fctr0:carrier.fctrT-Mobile 20.83592
## cellular.fctr0:carrier.fctrUnknown 20.83592
## cellular.fctr0:carrier.fctrVerizon 20.83592
## cellular.fctr1:carrier.fctrNone 20.83592
## cellular.fctrUnknown:carrier.fctrNone 20.83592
## cellular.fctrUnknown:carrier.fctrOther 20.83592
## cellular.fctrUnknown:carrier.fctrSprint 20.83592
## cellular.fctrUnknown:carrier.fctrT-Mobile 20.83592
## cellular.fctrUnknown:carrier.fctrVerizon 20.83592
## color.fctrUnknown 18.42273
## color.fctrSpace Gray 18.13706
## color.fctrWhite 17.33879
## prdl.descr.my.fctriPad1#1 16.25510
## prdl.descr.my.fctriPad1#0 13.80447
## color.fctrGold 10.72201
## startprice 20.61655
## prdl.descr.my.fctrUnknown#1 26.79449
## prdl.descr.my.fctriPadmini3#0 55.84036
## prdl.descr.my.fctriPadmini#0 26.27649
## prdl.descr.my.fctriPad2#1 20.77997
## cellular.fctrUnknown:carrier.fctrUnknown 14.50858
## cellular.fctrUnknown 15.08667
## prdl.descr.my.fctriPad4#1 25.73375
## Final.glmnet.importance
## biddable 50.03034
## prdl.descr.my.fctriPadminiRetina#0 71.05228
## prdl.descr.my.fctriPad2#0 24.82202
## prdl.descr.my.fctriPadAir2#0 81.59513
## prdl.descr.my.fctriPadAir#0 59.18262
## prdl.descr.my.fctriPadAir2#1 73.99464
## cellular.fctr1:carrier.fctrT-Mobile 29.48547
## prdl.descr.my.fctriPad4#0 44.83535
## prdl.descr.my.fctriPad3#0 32.79243
## prdl.descr.my.fctriPadmini2#0 48.36224
## cellular.fctr1 22.02544
## storage.fctr64 10.95262
## cellular.fctr1:carrier.fctrVerizon 25.75668
## cellular.fctr1:carrier.fctrUnknown 22.15689
## cellular.fctr1:carrier.fctrOther 100.00000
## prdl.descr.my.fctriPadminiRetina#1 73.90447
## prdl.descr.my.fctriPadmini3#1 55.35880
## prdl.descr.my.fctriPad5#0 54.87531
## prdl.descr.my.fctriPadAir#1 41.70199
## prdl.descr.my.fctriPadmini2#1 38.74030
## cellular.fctr1:carrier.fctrSprint 33.69155
## prdl.descr.my.fctriPad3#1 32.00882
## condition.fctrNew other (see details) 28.31638
## prdl.descr.my.fctriPadmini#1 25.48652
## condition.fctrNew 24.89291
## condition.fctrManufacturer refurbished 24.41207
## .rnorm 22.04002
## cellular.fctr0:carrier.fctrNone 20.83592
## cellular.fctr0:carrier.fctrOther 20.83592
## cellular.fctr0:carrier.fctrSprint 20.83592
## cellular.fctr0:carrier.fctrT-Mobile 20.83592
## cellular.fctr0:carrier.fctrUnknown 20.83592
## cellular.fctr0:carrier.fctrVerizon 20.83592
## cellular.fctr1:carrier.fctrNone 20.83592
## cellular.fctrUnknown:carrier.fctrNone 20.83592
## cellular.fctrUnknown:carrier.fctrOther 20.83592
## cellular.fctrUnknown:carrier.fctrSprint 20.83592
## cellular.fctrUnknown:carrier.fctrT-Mobile 20.83592
## cellular.fctrUnknown:carrier.fctrVerizon 20.83592
## color.fctrUnknown 18.42273
## color.fctrSpace Gray 18.13706
## color.fctrWhite 17.33879
## prdl.descr.my.fctriPad1#1 16.25510
## prdl.descr.my.fctriPad1#0 13.80447
## color.fctrGold 10.72201
## startprice 20.61655
## prdl.descr.my.fctrUnknown#1 26.79449
## prdl.descr.my.fctriPadmini3#0 55.84036
## prdl.descr.my.fctriPadmini#0 26.27649
## prdl.descr.my.fctriPad2#1 20.77997
## cellular.fctrUnknown:carrier.fctrUnknown 14.50858
## cellular.fctrUnknown 15.08667
## prdl.descr.my.fctriPad4#1 25.73375print("glb_newobs_df prediction stats:")## [1] "glb_newobs_df prediction stats:"if (glb_is_regression)
print(myplot_histogram(glb_newobs_df, paste0(glb_rsp_var_out, glb_fin_mdl_id)))
if (glb_is_classification)
print(table(glb_newobs_df[, paste0(glb_rsp_var_out, glb_fin_mdl_id)]))##
## N Y
## 446 352# Use this to see how prediction changes by changing one or more values
# players_df <- data.frame(id=c("Chavez", "Giambi", "Menechino", "Myers", "Pena"),
# OBP=c(0.338, 0.391, 0.369, 0.313, 0.361),
# SLG=c(0.540, 0.450, 0.374, 0.447, 0.500),
# cost=c(1400000, 1065000, 295000, 800000, 300000))
# players_df$RS.predict <- predict(glb_models_lst[[csm_mdl_id]], players_df)
# print(orderBy(~ -RS.predict, players_df))
# dsp_chisq.test(Headline.contains="[Vi]deo")
if ((length(diff <- setdiff(names(glb_trnobs_df), names(glb_allobs_df))) > 0) ||
(length(diff <- setdiff(names(glb_fitobs_df), names(glb_allobs_df))) > 0) ||
(length(diff <- setdiff(names(glb_OOBobs_df), names(glb_allobs_df))) > 0) ||
(length(diff <- setdiff(names(glb_newobs_df), names(glb_allobs_df))) > 0)) {
print(diff)
stop("glb_*obs_df not in sync")
}
if (glb_save_envir)
save(glb_feats_df, glb_allobs_df,
#glb_trnobs_df, glb_fitobs_df, glb_OOBobs_df, glb_newobs_df,
glb_models_df, dsp_models_df, glb_models_lst, glb_model_type,
glb_sel_mdl, glb_sel_mdl_id,
glb_fin_mdl, glb_fin_mdl_id,
file=paste0(glb_out_pfx, "prdnew_dsk.RData"))
sav_fin_mdl <- glb_fin_mdl; sav_sel_mdl <- glb_sel_mdl
save(sav_fin_mdl, sav_sel_mdl, file=paste0(glb_out_pfx, "sav_mdl.RData"))
# load(file=paste0(glb_out_pfx, "sav_mdl_01.RData"), verbose=TRUE)
# prv_fin_mdl <- sav_fin_mdl; prv_sel_mdl <- sav_sel_mdl
# load(file=paste0(glb_out_pfx, "sav_mdl.RData"), verbose=TRUE)
# cur_fin_mdl <- sav_fin_mdl; cur_sel_mdl <- sav_sel_mdl
# all.equal(cur_fin_mdl, prv_fin_mdl)
# cur_fitobs_df <- cur_fin_mdl$trainingData; prv_fitobs_df <- prv_fin_mdl$trainingData; all.equal(cur_fitobs_df, prv_fitobs_df)
# nrow(cur_fitobs_df); nrow(prv_fitobs_df)
# names(cur_fitobs_df); names(prv_fitobs_df)
# all.equal(cur_fin_mdl$bestTune, prv_fin_mdl$bestTune)
# all.equal(glb_sel_mdl, sav_sel_mdl)
# cur_fitobs_df <- cur_sel_mdl$trainingData; prv_fitobs_df <- prv_sel_mdl$trainingData; all.equal(cur_fitobs_df, prv_fitobs_df)
# head(myget_feats_importance(glb_sel_mdl)); head(myget_feats_importance(sav_sel_mdl))
# head(myget_feats_importance(cur_sel_mdl)); head(myget_feats_importance(prv_sel_mdl))
# tmp_replay_lst <- replay.petrisim(pn=glb_analytics_pn,
# replay.trans=(glb_analytics_avl_objs <- c(glb_analytics_avl_objs,
# "data.new.prediction")), flip_coord=TRUE)
# print(ggplot.petrinet(tmp_replay_lst[["pn"]]) + coord_flip())
glb_chunks_df <- myadd_chunk(glb_chunks_df, "display.session.info", major.inc=TRUE)## label step_major step_minor label_minor bgn end
## 16 predict.data.new 9 0 0 230.389 276.84
## 17 display.session.info 10 0 0 276.840 NA
## elapsed
## 16 46.451
## 17 NANull Hypothesis (\(\sf{H_{0}}\)): mpg is not impacted by am_fctr.
The variance by am_fctr appears to be independent. #{r q1, cache=FALSE} # print(t.test(subset(cars_df, am_fctr == "automatic")$mpg, # subset(cars_df, am_fctr == "manual")$mpg, # var.equal=FALSE)$conf) # We reject the null hypothesis i.e. we have evidence to conclude that am_fctr impacts mpg (95% confidence). Manual transmission is better for miles per gallon versus automatic transmission.
## label step_major step_minor label_minor bgn
## 11 fit.models 7 1 1 110.358
## 10 fit.models 7 0 0 52.896
## 16 predict.data.new 9 0 0 230.389
## 12 fit.models 7 2 2 180.214
## 9 select.features 6 0 0 30.402
## 14 fit.data.training 8 0 0 210.916
## 1 import.data 1 0 0 11.272
## 15 fit.data.training 8 1 1 223.102
## 13 fit.models 7 3 3 205.309
## 2 inspect.data 2 0 0 22.338
## 5 extract.features 3 0 0 27.765
## 3 scrub.data 2 1 1 26.487
## 8 partition.data.training 5 0 0 29.624
## 6 cluster.data 4 0 0 29.087
## 7 manage.missing.data 4 1 1 29.530
## 4 transform.data 2 2 2 27.673
## end elapsed duration
## 11 180.213 69.855 69.855
## 10 110.358 57.462 57.462
## 16 276.840 46.451 46.451
## 12 205.309 25.095 25.095
## 9 52.896 22.494 22.494
## 14 223.101 12.185 12.185
## 1 22.337 11.065 11.065
## 15 230.388 7.286 7.286
## 13 210.915 5.606 5.606
## 2 26.486 4.149 4.148
## 5 29.087 1.322 1.322
## 3 27.672 1.185 1.185
## 8 30.402 0.778 0.778
## 6 29.529 0.442 0.442
## 7 29.623 0.093 0.093
## 4 27.765 0.092 0.092
## [1] "Total Elapsed Time: 276.84 secs"